Jackendoff Ray: SEMANTIC INTERPRETATION IN GENERATIVE GRAMMAR (MIT
Press, 1972)
One of the milestone works in government and binding theory.
The author shows that theta roles determine to some extent the
wellformedness of anaphoric relations. Theta roles form a hierarchy
and binding must respect such hierarchy by placing the antecedent of
an anaphor higher on the hierarchy than the anaphor itself.
Jackendoff Ray: X'SYNTAX (MIT Press, 1977)(MIT Press, 1972)
A monumental study of the phrase structure of the english language in
the light of Chomsky's X-bar theory.
Jackendorff Ray: SEMANTICS AND COGNITION (MIT Press, 1983)
Jackendorff develops conceptual structures to explain language, in a
fashion similar to Fodor's mentalese.
The structure of meaning ought to be pursued on the same first princi-
ples as phonology and syntax.
Meaning of verbs can be reduced to a few spacetime primitives, such as
motion and location.
The "extended standard theory" enhances Chomsky's standard theory by
using interpretation rules to extract the meaning of a sentence. Such
rules apply to the intermediate syntactic structures used in the
derivation of the phonetic representation.
Jackendoff Ray: CONSCIOUSNESS AND THE COMPUTATIONAL MIND (MIT Press,
1987)
Jackendorff believes in a hierarchy of levels of mental representa-
tion.
The book resumes Jackendorff's claim that phonology and syntax are key
to the structure of meaning, then extends the framework developed for
language to vision and music (hinting at a possible unification with
Marr's theory of vision).
Each cognitive function exists at different levels of interpretations
and cognitive functions generally interact at intermediary levels.
Jackndorff refines and extends Fodor's idea of the modularity of the
mind.
Consciousness arises from a level of representation which is inter-
mediate between the sense-data and the form of thought.
Jackendorff Ray: SEMANTIC STRUCTURES (MIT Press, 1990)
Jackendorff's conceptual semantics is applied to lexical and syntactic
expressions in English. Jackendorff proposes a formalism for describ-
ing lexical semantic facts and expressing semantic generalizations.
He employs multi-dimensional representations analogous to those found
in phonology.
Jackendorff Ray: LANGUAGES OF THE MIND (MIT Press, 1992)
This collection of papers summarizes Jackendorff's formal theory on the nature of language and a modular approach to "mental anatomy", and applies the same concepts to learning and common sense reasoning. There is a tight relationship between vision and language. A lexical item contains the stereotipical image of the object or concept. Know- ing the meaning of a word implies knowing how the object or concept looks like.
Jackendoff Ray: PATTERNS IN THE MIND (Harvester Wheatsheaf, 1993)
Following Chomsky, Jackendoff thinks that the human brain contains
innate linguistic knowledge and that the same argument can be extended
to all facets of human experience: all experience is constructed by
unconscious genetically determined principles that operate in the
brain.
The experience of spoken language is constructed by the hearer's men-
tal grammar: speech per se is only a meaningless sound wave, only a
hearer equipped with the proper device can make sense of it.
These same conclusions can be applied to thought itself, i.e. to the
task of building concepts. Concepts are constructed by using some
innate, genetically determined, machinery, a sort of "universal gram-
mar of concepts". Language is but one aspect of a broader charac-
teristic of the human brain.
Jackson Frank: CONDITIONALS (Basil Blackwell, 1987)
A collection of articles by David Lewis, Robert Stalnaker, Grice and
Frank Jackson on the subject of conditionals. A theory of conditionals
must offer an account of the truth conditions of a conditional (under
which conditions "if A then B" is true or false, or acceptable to some
degree). The traditional view that a conditional is true if and only
if the antecedent is false or the consequent is true is too simplicis-
tic and allows conditionals such as "if Jones lives in London, then he
lives in Scotland" to be true (if he does not live in London or lives
in Scotland) when it is obviously senseless.
Stalnaker and Lewis solve some of the problems of (subjective) condi-
tionals ("if it were that A then it would be that B") by using
possible-world semantics. Lewis also reviews Ernest Adams' thesis that
the assertability of (indicative) conditionals ("if A then B") is
measured by the conditional probability of the consequent given the
antecedent.
Jackson Frank: PERCEPTION (Cambridge University Press, 1977)
The immediate objects of perception are mental. To perceive an object
is to be in a perceptual state as a causal result of the action of
that object.
On epiphenomenal qualia Jackson proposed a famous thought experiement:
a blind neurophysiologist that knows everything of how the brain per-
ceives colors still cannot know what it feels like to see a color.
Color is not a property of material things. Sense-data are not
material, they are mental.
Jauregui Jose: THE EMOTIONAL COMPUTER (Blackwell, 1995)
This is the english translation of 1990's "El Ordenador Cerebral".
Jauregi, like Wilson, views sociology as a branch of biology. The
same emotional system controls social, sexual and individual behavior.
Such emotional system originates from the neural organization of the
brain: emotions are rational and predictable events. Jauregi believes
that the brain is a computer, but introduced the novelty of emotions
as the direct product of that computer's processing activity. It is
emotions, not reason, that directs and informs the daily actions of
individuals. Jauregi deals with humans that feel pleasure and pain
rather than with abstract problem solvers.
Jauregi begins by separating the brain and the self: the brain is
aware of what is going on in the digestive system of the body, but
will inform the self only when some correction/action is necessary.
Normally, an individual is not aware of her digestive processes. Her
brain is always informed, though. The communication channel between
the brain and the self is made of emotions. The brain can tune the
importance of the message by controlling the intensity of the emo-
tions. Far from being an irrational process, the emotional life is
mathematically calculated to achieve exactly the level of response
needed. Feelings are subjective and inaccessible, but they also are
objective and precise.
The self has no idea of the detailed process that was going on in the
body and of the reason why that process must be corrected. The
brain's emotional system, on the other hand, is a sophisticated and
complex information-processing system. The brain is a computer pro-
grammed to inform the self (through emotions) of what must be done to
preserve her body and her society. It is through emotions that the
brain informs the self of every single detail in the body that is
relevant for survival. There almost is no instant without an emotion
that tells the individual to do something rather than something else.
"For human beings the reality that ultimately matters is the reality
of their feelings".
The self keeps a level of freedom: while it cannot suppress the (emo-
tional) messages it receives from the brain, it can disobey them. The
brain may increase the intensity of the message as the self disobeys
it a painful conflict may arise. The brain and the self are not only
separate, but they may fight each other.
Only the self can be conscious and feel, but the brain has control of
both consciousness and feelings.
If we view the brain as a computer, the hardware is made of the neural
organization. There are two types of software, though: bionatural
(knowledge about the natural world) and biocultural (such as a
language or a religion). A program has three main components: the
sensory, the mental and the emotional systems. Any sensory input can
be translated automatically by the brain into a mental (idea) or emo-
tional (feeling) message; and viceversa. Biocultural and bionatural
programs exhert emotional control over the body.
Jauregi distinguishes five systems of communication: the natural sys-
tem (the sender is a natural thing, such as a tree), the cultural sys-
tem (the sender is culture, something created by humans), the somatic
system (the sender is the individual's own body), the imaginary system
(the sender is imagination) and the social system (the sender is
another individual). The human brain is genetically equipped to
receive and understand all five kinds of messages. What ultimately
matters is the emotional translations of sensory inputs.
Jaynes Julian: THE ORIGIN OF CONSCIOUSNESS IN THE BREAKDOWN OF THE
BICAMERAL MIND (Houghton Mifflin, 1977)
Jaynes makes a number of interesting points about consciousness.
Consciousness is not necessary for concepts, learning, reason or even
thinking. Awareness of an action tends to follow, not precede, the
action. Awareness of an action bears little or no influence on the
outcome. Before one utters a sentence, one is not conscious of being
about to utter those specific words.
Consciousness is an operation rather than a thing. It is an operation
of analogy that transforms things of the real world into meanings in a
metaphorical space. Consciousness is a metaphor-generated model of the
world. Consciousness is based on language, therefore it appeared after
the emergence of language. By reviewing historical documents of past
civilizations, Jaynes tries to identify when and how consciousness was
born. Causes include the advent of writing, the loss of belief in
gods, epics, and natural selection itself.
Jaynes thinks that some social institutions and religions, psychologi-
cal phenomena such as hypnosis and schizophrenia, and artistic prac-
tices such as poetry and music are vestiges of an earlier stage of
human consciousness.
Jeanerrod Marc: THE COGNITIVE NEUROSCIENCE OF ACTION (Blackwell,
1996)
A survey of findings on the representations and processing that lead
to action, from neurophysiological data to the role of mental imagery.
Johnson-Laird Philip: HUMAN AND MACHINE THINKING (Lawrence Erlbaum,
1993)
A theory of deduction, induction and creation.
Johnson-Laird Philip: THINKING (Cambridge Univ Press, 1977)
A collection of articles that reviews the study of thinking in the
aftermath of the conceptual revolution that forced the transition from
behaviorism to information-processing. Contributions range from philo-
sophy (Popper, Kuhn) to artificial intelligence (Minsky, Schank).
Johnson-Laird Philip: MENTAL MODELS (Harvard Univ Press, 1983)
Johnson-Laird's representational theory assumes that mind represents
and processes models of the world. The mind solves problems without
any need to use logical reasoning. A linguistic representation such as
Fodor's is not necessary.
A sentence is a procedure to build, modify, extend a mental model.
The mental model created by a discourse exhibits a structure that
corresponds directly to the structure of the world described by the
discourse.
To perform an inference on a problem the mind needs to build the
situation described by its premises. Such mental model simplifies
reality and allows the mind to find an "adequate" solution.
Johnson-Laird draws on several phenomena to prove the psychological
inadequacy of a mental logic. People often make mistakes with deduc-
tive inference because it is not a natural way of thinking. The
natural way is to construct mental models of the premises: a model of
discourse has a structure that corresponds directly to the structure
of the state of affairs that the discourse describes. How can chil-
dren acquire inferential capabilities before they have any inferential
capabilities? Children solve problems by building mental models that
are more and more complex.
Johnson-Laird admits three types of representation: "propositions"
(which represent the world through sequences of symbols), "mental
models" (which are structurally analogous to the world) and "images"
(which are perceptive correlates of models).
Images are ways to approach models. They represent the perceivable
features of the corresponding objects in the real world.
Models, images and propositions are functionally and structurally dif-
ferent.
Linguistic expressions are first transformed into propositional
representations. The semantics of the mental language then creates
correspondences between propositional representations and mental
models, i.e. propositional representations are interpreted in mental
models.
Turning to meaning and model-theoretic semantics, Johnson-Laird pro-
poses that a mental model is a single representative sample from the
set of models satisfying the assertion. Semantic properties of expres-
sions are emergent properties of the truth conditions. Johnson-Laird's
procedural semantics assumes that there are procedures that construct
models on the basis of the meaning of expressions.
Johnson-Laird believes that consciousness is computable. The mind con-
tains a high-level operating system and a hierarchy of parallel pro-
cessors. Conscious mind is due to a serial process of symbolic mani-
pulation that occurs at the higher level of the hierarchy of proces-
sors (in the operating system), while unconscious mind is due to a
parallel process of distributed symbolic representation. Emotions are
non-symbolic signals, caused by cognitive interpretations of the
situation, that propagate within the hierarchy.
Johnson-Laird Philip: THE COMPUTER AND THE MIND (Harvard Univ Press,
1988)
An introduction to the themes and methods of cognitive science, with a
review of porduction and connectionist architectures. Speech, vision
and language are devoted long chapters. Johnson-Laird also introduces
his theory of mental models and resumes his theory of consciousness
and emotions.
Johnson-Laird Philip & Byrne Ruth: DEDUCTION (Lawrence Erlbaum, 1991)
The authors advance a comprehensive theory to explain all the main
varieties of deduction: propositional reasoning (that uses the connec-
tives "and", "or" and "not"), relational reasoning (that depends on
relations between entities), quantificational reasoning (that uses
quantifiers such as "any" and "some"). And justify it with a variety
of psychological experiments.
In order to understand discourse, humans construct an internal
representation of the state of affairs that is described in that
discourse. These mental models have the same structure as human con-
ceptions of the situations they represent. Deduction does not depend
on formal rules of inference but rather on a search for alternative
models of the premises that would refute a putative conclusion. Cen-
tral to the theory is the principle that people use models that make
explicit as little information as possible. The theory also make
sense of how people deal with conditionals.
The theory explains phenomena such as: that modus ponens ("if p then
q" and "p" then "q") is easier than modus tollens ("if p then q" and
"not q" then "not p").
Josephson John & Josephson Susan: ABDUCTIVE INFERENCE (Cambridge
University Press, 1993)
Abduction (inference to the best explanation, i.e. building the
hypothesis that best accounts for the data) is ubiquitous in ordinary
life as well as in scientific theory formation. The book presents a
dynasty of systems that explored abduction. Intelligence is viewed as
a cooperative community of knowledge-based specialists (performing
"generic tasks"). Knowledge arises from experience by processes of
abductive inference.
Jouvet Michel: LE SOMMEIL ET LE REVE (Jacob, 1992)
Jouvet was the first to localize the trigger zone for REM sleep and
dreaming in the brain stem. In this book he provides a neurobiological
and psychological analysis of sleep and dreaming.
According to his findings, a dream is the vehicle employed by an
organism to cancel or archive the day's experiences on the basis of a
genetic program. Dreaming is a process that absorbs a lot of energy.
This theory would also solve the dualism between hereditary and
acquired features. An hereditary component is activated daily to
decide how new data must be acquired.
Kaku Michio: HYPERSPACE (Oxford University Press, 1994)
A popular introduction to modern cosmology, including black hole,
time travel, parallel universes and alien civilizations.
The title refers to the fact that the universe may actually exist
in dimensions beyond the commonly accepted four of spacetime. The
laws of nature become simpler when expressed in higher dimen-
sions. In fact, all forces can be unified in the ten-dimensional
hyperspace of superstring theory. Kaku shows how the concept of
supergravity was derived from the intuitions of the old Kaluza-
Klein theory, which first unified the two great field theories,
light and gravity (Maxwell and Einstein).
Kandell Abraham: FUZZY MATHEMATICAL TECHNIQUES (Addison Wesley,
1986)
A very technical and very well organized introduction to the con-
cepts and theorems of fuzzy logic: fuzzy sets, theory of possi-
bility, fuzzy functions (integration and differentiation), mul-
tivalent logics, linguistic approximation and applications.
Kanerva Pentti: SPARSE DISTRIBUTED MEMORY (MIT Press, 1988)
The sparse distributed memory is a model of long-term memory in
which situations are encoded by patterns of features and episodes
are encoded by sequences of them. Any pattern in a sequence can
be used to retrieve the entire sequence. Memories are stored
based on features. The senses must extract the invariant features
of objects to retrieve the corresponding memories. The motor sys-
tem is also controlled by sequences of patterns in memory. A cen-
tral site, the "focus", stores all the features that are needed
to define the specific moment in time, to account for subjective
experience. The model is capable of learning.
Most of the study is a computational analysis of the feasibility
of a very large address space whose units of address decoding are
linear threshold functions (neurons).
Kaplan David: THEMES FROM KAPLAN (Oxford Univ Press, 1989)
This book is a tribute to Kaplan by a number of thinkers (Cas-
taneda, Church, Deutsch, etc), but also contains Kaplan's famous
"Demonstratives" (1977).
Indexicals include the personal pronouns, the demonstrative pro-
nouns, some adverbs ("here", "now", "tomorrow"), etc, i.e. words
whose referent depends on the context of use (whose meaning pro-
vides a rule which determines the referent in terms of the con-
text). The logic of demonstratives, based on first-order predi-
cate logic, is a theory of word meaning, not speaker's meaning,
based on linguistic rules shared by all linguistic users.
Indexicals are "directly referential", i.e. refer directly to
individuals without the mediation of Fregean sense (unlike nonin-
dexical definite descriptions, which denote their referent
through their sense). Kaplan's indexicals are similar to
Kripke's "rigid designators", expressions that designate the
same thing in every possible world in which they exist and desig-
nate nothing elsewhere. Indexicals provide directly that the
referent in every circumstance is fixed to be the actual
referent. In Kaplan's case, though, the expression is the "dev-
ice" of direct reference.
Kaplan distinguishes between the "character" of a linguistic
expression (its grammatical meaning, i.e. what the hearer learns
when she learns the meaning of that expression) and its "content"
in a context (the proposition, the primary bearer of truth-
values, the object of thought). Indexicals have a context-
sensitive character, nonindexicals have a fixed character. Char-
acters are functions that map contexts into contents.
The theory of direct reference for indexicals includes: the
language system (to which meanings and characters belong), the
contexts of uses (through which referents are assigned to expres-
sions) and the circumstances of evaluation (at which truth-values
are allocated to sentential referents).
Karmiloff-Smith Annette: BEYOND MODULARITY (MIT Press, 1992)
A developmental model is proposed that embraces both Piaget's
constructivism and Fodor's nativism, both innate capacities of
the human mind and subsequent representational changes. Based on
a number of experiments on children, Karmiloff-Smith believes
that initially children learn by instinct, or at least impli-
citly; then their thinking develops, by redescribing the world
from an implicit form to more and more explicit forms, to more
and more verbal knowledge. She contends that there are separate
cognitive categories, each with its own innate structure; but
modularization is seen as a product of the child's development
and development proceeds along the same sequential steps for all
mental activities. Language is just one of them. The model is
then applied to connectionist models of the mind.
Katz Jerrold: THE METAPHYSICS OF MEANING (MIT Press, 1990)
A critique of naturalism, particularly Wittgenstein's argument
against intensionalist theories of meaning and Quine's argument
for indeterminacy. By examining Wittgenstein's own critique of
pre-existing theories of meaning, Katz salvages a theory of mean-
ing (the "proto-theory") which postulates underlying sense struc-
ture (just like Chomsky's postulation of underlying syntactic
structure) and constructs a decompositional semantics (i.e., pro-
vides a preliminary theory of decompositional sense structure).
Katz replaces Frege's referentially defined notion of sense with
a notion defined in terms of sense properties and relations
internal to the grammar of the language, thereby accomplishing a
separation of sense structure and logical structure (a separation
of grammatical meaning from reference and use).
Katz thinks that words' meaning can be decomposed in atoms of
meaning that are universal for all languages.
This may well be the most detailed critique ever of
Wittgenstein's thought.
Katz Jerrold: AN INTEGRATED THEORY OF LINGUISTIC DESCRIPTIONS
(MIT Press, 1964)
Two components are necessary for a theory of semantics: a dic-
tionary, which provides for every lexical item a phonological
description, a syntactic classification ("grammatical marker",
e.g. noun or verb) and a specification of its possible distinct
senses ("semantic marker", e.g. light as in color and light as
the opposite of heavy); and projection rules, which produce all
valid interpretations of a sentence.
Katz Jerrold: THE PHILOSOPHY OF LANGUAGE (Harper & Row, 1966)
According to Katz, a theory of language is a theory of linguistic
universals (features that all languages have in common). Katz
argues that the basic ontological categories are those semantic
markers that are implied by other semantic markers but never
imply other markers themselves.
Katz Jerrold: SEMANTIC THEORY (Harper & Row, 1972)
Two components are necessary for a theory of semantics: a dic-
tionary, which provides for every lexical item a phonological
description, a syntactic classification ("grammatical marker",
e.g. noun or verb) and a specification of its possible distinct
senses ("semantic marker", e.g. light as in color and light as
the opposite of heavy); and projection rules, which produce all
valid interpretations of a sentence.
"The logical form of a sentence is identical with its meaning as
determined compositionally from the senses of its lexical items
and the grammatical relations between its syntactic consti-
tuents."
Kaufmann Arnold & Gupta Madan: INTRODUCTION TO FUZZY ARITHMETICS
(Van Nostrand Reinhold)
A technical (and one of the most rigorous) introduction to the
properties of fuzzy numbers. A fuzzy number is viewed as an
extension of an interval of confidences, once it is related to a
level of presumption. The addition of fuzzy numbers and random
data yields hybrid numbers, which transform a measurement of an
objective data into a valuation of a subjective value without any
loss of information. Definitions are provided for derivatives of
functions of fuzzy numbers, fuzzy trigonometric functions, etc.
Kauffman Stuart: THE ORIGINS OF ORDER (Oxford University Press,
1993)
Darwin's vision of natural selection as a creator of order is not
sufficient to explain all the spontaneous order exhibited by both
the living and the dead universe. At every level of science the
spontaneous emergence of order, or self-organization of complex
systems, is a common theme.
Living organisms happen to be mere accidents in this universal
process. Natural selection and self-organization complement each
other: they create complex systems poised at the edge between
order and chaos, which are fit to evolve in a complex environ-
ment. The target of selection is a type of adaptive system at the
edge between chaos and order. This is one of the three types of
behaviors that are possible for large networks of elements
(besides chaotic and ordered). This applies at all levels of
organization, from living organisms to ecosystems.
Kauffman's mathematical model involves "fitness landscapes". A
fitness landscape is a distribution of fitness values over the
space of genotypes. Adaptive evolution can be represented as a
local hill climbing search converging via fitter mutants toward
some local or global optimum. Adaptive evolution occurs on
rugged (multipeaked) fitness landscapes. The very structure of
these landscapes implies that radiation and stasis are inherent
features of adaptation. The Cambrian explosion and the Permian
extinction may be the natural consequences of inherent properties
of rugged landscapes.
Kauffman also advances his theory of how life may have ori-
ginated. When a system of simple chemicals reaches a certain
level of complexity, it undergoes a phase transition. The
molecules spontaneously combine to yield larger molecules of
increasing complexity and catalytic capability. Such autocata-
lytic chemical processes may have formed the basis for early
life. Life began complex, with a metabolic web which was capable
of capturing energy sources.
Arrays of interacting genes do not evolve randomly but converge
toward a relatively small number of patterns, or "attractors".
This ordering principle may have played a larger role than did
natural selection in guiding the evolution of life.
Principles of self-organization also drive the genetic program
which drives morphogenesis. A few behaviors at the cell level
(e.g., differentiation) are actually unavoidable consequences of
the properties of self-organization. They are not the product of
selection, but rather of the properties of the systems that
selection acts upon. Laws of form complement selection. In any
event, the genetic program is not a sequence of instructions but
rather a regulatory network that behaves, again, like a self-
organizing system.
Kauffman is searching for the fundamental force that counteracts
the universal drift towards disorder required by the second law
of thermodynamics.
Kauffman Stuart: AT HOME IN THE UNIVERSE (Oxford Univ Press,
1995)
The whole is greater than its parts: life is not located in any
of the parts of a living organism, but arises from the emergent
properties of the whole they compose. Such emergent properties
are the result of a ubiquitous trend towards self-organization.
Self-organizing principles are inherent in our universe and life
is a direct consequence of self-organization. Therefore, both the
origin of life and its subsequent evolution were inevitable.
Kauffman refutes the theory that life started simple and became
complex in favor of a scenario in which life started complex and
whole due to a property of some complex chemical systems, the
self-sustaining process of autocatalytic metabolism. Life is but
a phase transition that occurs when the system becomes complex
enough. Life is vastly more probably than traditionally assumed.
The theme of science is order. Order can come from equilibrium
systems and from non-equilibrium systems that are sustained by a
constant source of matter/energy or (udally) by a persistent dis-
sipation of matter/energy. In the latter systems, order is gen-
erated by the flux of matter/energy. All living organisms (as
well as systems such as the biosphere) are nonequilibrium ordered
systems.
Kauffman advocates a "theory of emergence" that deals with none-
quilibrium ordered systems. Such a theory would explain why life
emerged at all.
Evolution is viewed as the traversing of a fitness landscape.
Peaks represent optimal fitness. Populations wander driven by
mutation, selection and drift across the landscape in their
search for peaks. It turns out that the best strategy for reach-
ing the peaks occurs at the phase transition between order and
disorder (the "edge of chaos"). The same model applies to other
biological phenomena and even nonbiological phenomena, and may
therefore represent a universal law of nature.
Kauffman's view of life can be summarized as: autocatalytic net-
works arise spontaneously; natural selection brings them to the
edge of chaos; a genetic regulatory mechanism accounts for onto-
geny. Natural selection is not the only source of order: there is
also order for free.
The main theme of Kauffman's research is that the requirements
for order to emerge are far easier than traditionally assumed
("order for free").
Kavanaugh Robert: EMOTION (Lawrence Erlbaum, 1996)
An overview of studies on emotion.
Kaye Jonathan: PHONOLOGY (Lawrence Erlbaum, 1989)
A cognitive approach to phonology. Besides reviewing the history
of the field and the recent developments (syllable structure,
tones and nonlinear phonology, harmony, parametrized systems),
Kaye advances his own theory that the function of phonological
processes is to help process language in a fashion similar to
punctuation by providing information about domain boundaries. A
theory of markedness was also sketched to explain the fact that
certain features condition other features.
Kearns Michael & Varizani Umesh: INTRODUCTION TO COMPUTATIONAL
LEARNING THEORY (MIT Press, 1994)
A very technical survey of the main issues of learning theory,
built around Valiant's "probably approximately correct" model
(1992), which defines learning in terms of the predictive power
of the hypothesis output by the learning algorithm. Notions such
as the Vapnik & Chervonenkis dimension, a measure of the sample
complexity of learning, and various extensions to Valiant's algo-
rithm are presented.
Keil Frank: SEMANTIC AND CONCEPTUAL DEVELOPMENT (Harvard Univ
Press, 1979)
Following Fred Sommers, Keil develops a formal theory of the
innate constraints that guide and limit the acquisition of onto-
logical knowledge (knowledge about the basic categories of the
world). Two terms are of the same type if all predicates that
span one of them also span the other one; and two predicates are
of the same type if they span exactly the same sets of terms. No
two terms have intersecting predicates. No two predicates span
intersecting sets of terms (the "M constraint"). Ontological
knowledge is therefore organized in a rigid hierarchical fashion.
Keil Frank: CONCEPTS, KINDS AND COGNITIVE DEVELOPMENT ( (Cam-
bridge University Press, 1989)
Concepts are always related to other concepts. No concept can be
understood in isolation from all other concepts. Concepts are not
simple sets of features. Concepts embody "systematic sets of
causal beliefs" about the world and contain implicit explanations
about the world. Concepts are embedded in theories about the
world, and they can only be understood in the context of such
theories.
In contrast with stage-based developmental theories, Keil argues
for the continuity of cognition across development. Continuity is
enforced by native constraints on developmental directions.
Perceptual procedures through which objects are categorized are
not part of the categories: an animal is a skunk if its mother is
a skunk regardless of what it looks like.
Keil refines Quine's ideas. Natural kinds are not defined by a
set of features or by a prototype: they derive their concept from
the causal structure that underlies them and explains their
superficial features. They are defined by a "causal homeostatic
system", which tends to stability over time in order to maximize
categorizing. Nominal kinds (e.g., "odd numbers") and artifacts
(e.g., "cars") are similarly defined by the theories they are
embedded in, although such theories are qualitatively different.
There is a continuum between pure nominal kinds and pure natural
kinds with increasing well-definedness as we move towards natural
kinds. What develops over time is the awareness of the network of
causal relations and mechanisms that are responsible for a
natural kind's essential properties. The theory explaining a
natural kind gets refined over the years.
Kelso Scott & Mandell Arnold: DYNAMIC PATTERNS IN COMPLEX SYS-
TEMS (World Scientific, 1988)
Proceedings of a 1988 conference on self-organizing systems.
Hermann Haken discusses the dualism between pattern recognition
and pattern formation.
Kelso shows that the brain exhibits processes of self-
organization that obey to nonlinear dynamics features (multista-
bility, abrupt phase transitions, crises and intermittency). The
human behavior is therefore also subject to nonlinear dynamics.
Kelso Scott: DYNAMIC PATTERNS (MIT Press, 1995)
Kelso believes that all levels of behavior, from neural processes
to mind, are governed by laws of self-organization. He explains
human behavior from phenomena of multistability, phase transi-
tions, etc.
Kessel Frank: SELF AND CONSCIOUSNESS (Lawrence Erlbaum, 1993)
A collection of essays on the subject, with contributions by Den-
nett, Neisser and Gazzaniga.
Kim Jaegwon: SUPERVENIENCE AND MIND (Cambridge University Press,
1993)
A collection of philosophical essays, particularly on superveni-
ence.
The world has a structure: the existence of an object and its
properties depend on, or are determined by, the existence and the
properties of other objects. With Hume, "causation is the cement
of the universe". Supervenience is a type of relation between
objects that occurs between their properties: if two individuals
are alike in all their physical properties, then they must be
alike also in their nonphysical properties, i.e. the set of
valuational (nonphysical) properties supervenes on the set of
nonvaluational (physical) ones.
"Supervenience" theory assumes that objects with the same physi-
cal properties also exhibit the same mental properties. A causal
relation between two states can be explained both in mental terms
and in physical terms. The mental and the physical interact only
to guarantee consistence. The mental supervenes on the physical,
just like the macroscopic properties of objects supervene on
their microscopic structures.
In general, supervenience is a relation between two sets of pro-
perties over a single domain (e.g., mental and physical proper-
ties over the domain of organisms). Weak supervenience occurs
when indiscernibility with respect to a class of properties
entails indiscernibility with respect to another class of proper-
ties. Strong supervenience claims that if individuals share the
same physical properties, then they must share the same mental
properties. Global supervenience occurs when worlds that are
indiscernible with respect to an individual are also indiscerni-
ble with respect to another individual.
Kim is a physicalist (the world is a physical world governed by
physical laws) and a mental realist (mentality is a real feature
of the world and has the power to cause events of the world). His
goal is to understand how the mind can "cause" anything in the
physical world.
Kirkham Richard: THEORIES OF TRUTH (MIT Press, 1992)
A philosophical (and probably unique) introduction to a variety
of modern theories of truth: Charles Peirce's pragmaticism, Wil-
liam James' instrumentalism, Brand Blanshard's coherence theory
(truth as a fully coherent set of beliefs), Russell's congruence
theory,nd theory of types Austin's correlation theory, Tarski's
correspondence theory. Theories of justification (how to iden-
tify the properties of true statements by reference to which the
truth of a statement can be judged) are treated as separated from
theories of truth, as well as theories of speech acts. The sys-
tems of Davidson, Dummett, Kripke, Prior are reviewed and criti-
cized.
Kitchener Robert: PIAGET'S THEORY OF KNOWLEDGE (Yale University
Press, 1986)
One of the best introduction to genetic epistemology.
Kittay Eva: METAPHOR (Clarendon Press, 1987)
Drawing from Black's interactionist theory, and its vision of
metaphor's dual content (literal and metaphorical, "vehicle" and
"topic"), Kittay develops a theory of metaphor
Kittay's theory of metaphor is based on her own "relational"
theory of meaning, which is inspired by Saussure's theory of
signs. The meaning of a word is determined by other words that
are related to it by the lexicon. Meaning is not an item, is a
field. A semantic field is a group of words that are semantically
related to each other. Language is context-dependent, and con-
textual features are constitutive of meaning.
Metaphor is a process that transfers semantic structures between
two semantic fields: some structures of the first field creates
or reorganizes a structure in the second field.
The meaning of a word consists of all the literal senses of that
word. A literal sense consists of a conceptual content, a set of
conditions, or semantic combination rules (permissible semantic
combinations of the word, analogous to Fodor's selection-
restriction rules) and a semantic field indicator (relation of
the conceptual content to other concepts in a content domain).
An interpretation of an utterance is any of the senses of that
utterance. Projection rules combine lower-level units into
higher-level units according to their semantic combination rules.
A first-order interpretation of an utterance is derived from a
valid combination of the first-order meanings of its consti-
tuents. Second-order interpretation is a function of first-order
interpretation and expresses the intuitive fact that what has to
be communicated is not what is indicated by the utterance's
literal meaning.
Kittay outlines the formal conditions for recognizing an utter-
ance as a metaphor. An explicit cue to the metaphorical nature of
an utterance is when the first-order and the second-order
interpretation point to two distinct semantic fields.
Equivalenty, an incongruity principle (incongruity between a
focus and a frame) can be used. discriminate a metaphorical
utterance.
Metaphor can be interpreted as second-order meaning.
The cognitive force of metaphor comes from a reconceptualization
of information about the world that has already been acquired but
possibly not conceptualized. Metaphor turns out to be one of the
primary ways in which humans organize their experience.
Metaphorical meaning is not reducible to literal meaning.
Klahr David: PRODUCTION SYSTEM MODELS OF LEARNING AND DEVELOP-
MENT (MIT Press, 1987)
A set of articles that provide an overview of production systems
from the perspective of cognitive psychology and in the context
of working computer programs. Includes Pat Langley's "A general
theory of discrimination learning" (the PRISM project) and Paul
Rosenbloom's "Learning by chunking" (the XAPS project).
Kleene Stephen: INTRODUCTION TO METAMATHEMATICS (North-Holland,
1964)
Kleene's three-valued logic was conceived to accomodate undediced
mathematical statements. The third truth value signals a state of
partial ignorance. The undecided value is assigned to any well-
formed formula that has at least one undecided component.
Klopf Harry: THE HEDONISTIC NEURON (Hemisphere, 1982)
Organisms actively seek stimulation. If homeostasis is the seek-
ing of a steady-state condition, "heterostasis" is the seeking of
a maximum stimulation. All parts of the brain are independently
seeking positive stimulation (or "pleasure") and avoiding nega-
tive stimulation (or "pain"). All parts are goal-driven in that,
when responding to a given stimulus leads to "pleasure", the
brain part will respond more frequently to that stimulus in the
future; and viceversa.
In his neural model cognition and emotion cohexist and complement
each other. Emotion provides the sense of what organisms need.
Cognition provides the means for achieving those needs.
Kodratoff Yves: INTRODUCTION TO MACHINE LEARNING (Morgan Kauf-
man, 1988)
A technical, Prolog-oriented textbook on machine learning that
starts with the theoretical foundations of production systems,
deals with truth maintenance and then surveys a number of learn-
ing methods: Mitchell's version spaces, explanation-based (deduc-
tive) learning, analogical learning, clustering.
Koestler Arthur: THE GHOST IN THE MACHINE (Henry Regnery, 1967)
Koestler brings together a wealth of biological, physical,
anthropological and philosophical arguments to construct a uni-
fied theory of open hierarchical systems.
Language has to do with a hierarchic process of spelling out
implicit ideas in explicit terms by means of rules and feedbacks.
Organisms and societies also exhibit the same hierarchical struc-
ture. Each intermediary entity ("holon") function as self-
contained wholes relative to their subordinates and as dependent
parts to their superordinates. Each holon tends to persist and
assert its pattern of activity.
Wherever there is life, it must be hierarchically organized.
Life exhibits an integrative property (that manifests itself as
symbiosis) that enables the gradual construction of complex
hierarchies out of simple holons. In nature there are no
separated, indivisible, self-contained units. An "individual" is
an oxymoron. An organism is a hierarchy of self-regulating
holons (a "holarchy") that work in coordination with their
environment. Holons at the higher levels of the hierarchy have
progressively more degrees of freedom and holons at the lower
levels of the hierarchy have progressively less degrees of free-
dom. Moving up the hierarchy, we encounter more and more com-
plex, flexible and creative patterns of activity. Moving down the
hierarchy behavior becomes more and more mechanized.
A hierarchical process (which gradually reduces the percept to
its fundamental elements) is also involved in perception and
memorization. A dual hierarchical process (which gradually
reconstructs the percept) is involved in recalling.
Hierarchical processes of the same nature can be found in the
development of the embryo, in the evolution of species and in
consciousness itself (which should be analyzed not in the context
of the mind/body dichotomy but in the context of a multi-levelled
hierarchy and of degrees of consciousness).
They all share common themes: a tendency towards integration (a
force that is inherent in the concept of hierarchic order, even
if it seems to challenge the second law of thermodynamics as it
increases order), an openess at the top of the hierarchy (towards
higher and higher levels of complexity) and the possibility of
infinite regress.
Kohonen Teuvo: ASSOCIATIVE MEMORY (Springer Verlag, 1977)
The retrieval of information in memory occurs via associations.
An associative memory is a system from which a set of information
can be recalled by using any of its members. An adaptive associa-
tive network is viewed as a reasonable model for biological
memory. Kohonen also argues for the biological plausibility of
holographic associative memories. For each model a thorough
mathematical treatment is provided.
Kohonen Teuvo: SELF-ORGANIZATION AND ASSOCIATIVE MEMORY
(Springer Verlag, 1984)
A formal study of memory from a system theory's viewpoint.
Kohonen built a psychologically-plausible model of how the brain
represents topographically the world, with nearby units respond-
ing similarly. His model is therefore capable of self-organizing
in regions.
Kohonen's connectionist architecture, inspired by Malsburg's stu-
dies on self-organization of cells in the cerebral cortex, is
able to perform unsupervised training, i.e. it learns categories
by itself.
Instead of using Hebb's learning, Kohonen assumes that the
overall synaptic resources of a cell are approximately constant
and what changes is the relative efficacies of the synapses. A
neural network has learned a new concept when the weights of con-
nections converge towards a stable configuration. This model
exhibits mathematical properties that set it apart: the layering
of neurons plays a specific role (the wider the intermediate
layer, the faster but the more approximate the process of
categorization).
A variant of Hebb's law yields competitive behavior.
Kohonen also reviews classical learning systems (Adaline, Percep-
tron) and holographic memories.
Kohonen Teuvo: SELF-ORGANIZING MAPS (Springer Verlag, 1995)
The Adaptive-Subspace Self Organizing Map (ASSOM) is an algorithm
for neural networks that combines Learning Subspace Method (LSM),
the first supervised competitive-learning algorithm ever, and
Self Organizing Map (SOM), another algorithm invented by Kohonen,
that maps patterns close to each other in the input space onto
contiguous locations in the output space (topology preserving).
The new algorithm is capable of detecting invariant features.
Kolodner Janet & Riesbeck Christopher: EXPERIENCE, MEMORY, AND
REASONING (Lawrence Erlbaum, 1986)
An introduction to computational theories of memory that are
derived from the conceptual dependency theory. Each article is
written by an expert in the field. Schank writes about
explanation-based learning. Lebowitz describes his RESEARCHER
project. Lytinen discusses his word-based parsing technique.
Riesbeck introduces to his direct memory access parsing system.
Kolodner Janet: CASE-BASED REASONING (Morgan Kaufmann, 1993)
A monumental summary of the discipline of case-based systems that
also attempts ot lay logical foundations for the field. Emphasis
is placed on the views of learning as a by-product of reasoning,
and reasoning as remembering; on the essential task of adapting
old solutions to solve new problems (old cases to explain new
situations). Schank's cognitive model of dynamic memory (MOPs
and the likes) is introduced at length. Some of the historical
systems (CHEF, CYRUS, etc) are discussed. The book provides
detailed techniques for storing, indexing, retrieving, matching
and using cases.
Kolodner Janet: RETRIEVAL AND ORGANIZATIONAL STRATEGIES IN CON-
CEPTUAL MEMORY (Lawrence Erlbaum, 1984)
A description of the CYRUS system, which was based on Schank's
conceptual dependency theory.
Kosko Bart: NEURAL NETWORKS AND FUZZY SYSTEMS (Prentice Hall,
1992)
A textbook on adaptive fuzzy systems that presents a unified view
of neural networks and fuzzy systems. Kosko presents neural net-
works as stochastic gradient systems and fuzzy sets as points in
unit hypercubes.
All the main learning algorithms for neural networks are reviewed
and formalized. It is shawn that neural computations is similar
to statistics in that its goal is to approximate the function
that relates a set of inputs to a set of outputs.
In Kosko's formalization, a fuzzy set is a point in the unitary
hypercube equivalent to Zadeh's universe of discourse, and a
non-fuzzy set is one of the vertexes of such a cube. The para-
doxes of classical logic occur in the middle points of the hyper-
cube.
A fuzzy set's entropy (which could be thought of as its "ambi-
guity") is defined by the number of violations of the law of
non-contradiction compared with the number of violations of the
excluded middle. Entropy is zero when both laws hold, is maximum
in the center of the hypercube. Alternatively, a fuzzy set's
entropy can be defined as a measure of how a set is a subset of
itself.
A fuzzy system is a relationship between hypercubes, a relation-
ship of fuzzy sets into families of fuzzy sets.
Fuzzy associative memories are balls of fuzzy sets into balls of
fuzzy sets.
Fuzzy logic, which can account for all results of the theory of
probability, better represents the real world, without any need
to assume the existence of randomness. For example, relative fre-
quency is a measure of how a set is a subset of another set.
Many of Physics' laws are not reversible because if they were
casuality would be violated (after a transition of state proba-
bility turns into certainty and cannot be rebuilt working back-
wards). If they were expressed as "ambiguity", rather than proba-
bility, they would be reversible, as the ambiguity of an event
remains the same before and after the event occurred.
The space of neural states (the set of all possible outputs of a
neural net) is identical to the power fuzzy set (the set of all
fuzzy subsets of the set of neurons). A set of "n" neurons (whose
signals vary continously between zero and one) defines a family
of n-dimensional fuzzy sets. That space is the unitary hypercube,
the set of all vectors of length "n" and coordinates in the uni-
tary continous interval (zero to one).
Hopfield's nets tend to push the state of the system towards one
of the 2 to the "n" vertexes of the hypercube. This way they
dynamically disambiguate fuzzy descriptions by minimizing their
fuzzy entropy.
Kosko Bart: FUZZY THINKING (Hyperion, 1993)
Fuzziness is pervasive in nature ("everything is a matter of
degree"), while science does not admit fuzziness.
Even probability theory still assumes that properties are crisp.
And probability (according to Kosko's "subsethood" theorem) can
be interpreted as a measure of how much the whole (the space of
all events) is contained in the part (the event). Kosko shows
how logical paradoxes such as Russell's can be interpreted as
"half truths" in the context of fuzzy logic. Heisenberg's uncer-
tainty principle (the more a quantity is accurately determined,
the less accurately a conjugate quantity can be determined, which
holds for position and momentum, time and energy) can be reduced
to the Cauchy-Schwarz inequality (which is related to Pythagora's
theorem, which is in turn related to the subsethood theorem).
Applications such as fuzzy associative memories, adaptive fuzzy
systems and fuzzy cognitive maps are discussed at length.
Kosko even discusses why the universe exists (because otherwise
the fuzzy entropy theorem would exhibit a singularity) and specu-
lates that the universe is information and maybe God himself is
information.
Too much autobiography and too many references to eastern reli-
gion try to make the book more accessible but probably merely
detract from the subject.
Kosslyn Stephen: IMAGE AND MIND (Harvard University Press, 1980)
"Mental imagery" is seeing something in the absence of any sen-
sory signal, such as the perception of a memory. Kosslyn analyzes
what is seen when in the brain there is no such image, and why we
need mental imagery at all.
Based on numerous psychological experiments, Kosslin maintains
that mental imagery is pictorial in character, i.e. that mental
imagery involves scanning an internal picture-like entity. Mental
images can be inspected and classified using pretty much the same
processes used to inspect and classify visual perceptions.
To explain the structure of mental imagery Kosslyn puts forth a
representational theory of the mind of a "depictive" type, as
opposed to Fodor's propositional theory and related to Johnson-
Laird's models. Kosslyn thinks that the mind can build visual
representations, which are coded in parts of the brain, and which
reflect what they represent. Such representations can be
inspected by the mind and transformed (rotated, enlarged,
reduced).
There exist two levels of visual representation: a "geometric"
level, which allows one to mentally manipulate images, and an
"algebric" one, which allows one to "speak" about those images.
Kosslyn thinks that mental imagery achieves two goals: retrieve
properties of objects and predict what would happen if the body
or the objects should move in a given way. Reasoning on shapes
and dimensions is far faster when we employ mental images rather
than concepts.
Kosslyn Stephen: GHOSTS IN THE MIND'S MACHINE (W. Norton, 1983)
An introduction to Kosslyn's theory of mental imagery oriented
towards a computer implementation.
Kosslyn Stephen & Koenig Olivier: WET MIND (Free Press, 1992)
An overview of cognitive neuroscience, i.e. of psychological stu-
dies based on the principle that "the mind is what the brain
does", i.e. theories that describe mental events by means of
brain activities.
Chapters on neural computation, vision, language, movement,
memory.
Kosslyn Stephen: IMAGE AND BRAIN (MIT Press, 1994)
This book revises and expands the contents and conclusions of
"Image and Mind".
Kosslyn's proposal for the resolution of the imagery debate is an
interdisciplinary theory of high-level vision in which perception
and representation are inextricably linked. Visual perception
(visual object identification) and visual mental imagery share
common mechanisms.
Visual processing is decomposed in a number of subsystems, each a
neural network: visual buffer (located in the occipital lobe),
attention window (selects a pattern of activity in the visual
buffer), two cortical visual systems, the ventral system (infe-
rior temporal lobe, encodes object properties) and the dorsal
system (posterior paretal lobe, encodes spatial properties),
associative memory (which integrates the two classes of proper-
ties), information lookup subsystem (dorsolaterla prefrontal
cortex, accesses information about the most relevant object in
associative memory), attention shifting subsystems (frontal,
parietal and subcortical areas, directs the attention window to
the appropriate location). The subsystems may overlap and
exchange feedback. More detailed analysis of the visual recogni-
tion process identify more specialized subsystems. The model is
therefore gradually extended to take into account the full taxo-
nomiy of visual abilities.
Mental imagery shares most of this processing architecture with
high-level visual perception.
During the course of the development of the theory, a wealth of
psychological and neurophysiological findings is provided.
Koza John: GENETIC PROGRAMMING (MIT Press, 1992)
One of the seminal books on "genetic" programming by means of
natural selection. The solution to a problem is found by geneti-
cally breeding populations of computer programs. A computer is
therefore enabled to solve problems without being explicitly pro-
grammed to solve them. The process of finding a solution to a
problem is turned into the process of searching the space of com-
puter programs for a highly fit individual computer program to
solve such a problem.
Koza John: GENETIC PROGRAMMING II (MIT Press, 1994)
Focuses on automatic function definition for the decomposition of
complex problems.
Kripke Saul: NAMING AND NECESSITY (Harvard University Press,
1980)
Kripke developed a model-theoretic interpretation of various
axiom sets for modal logic. Modality can be represented by recur-
ring to the notion of possible worlds. In Kripke's semantics a
property is necessary if it is true in all worlds, a property is
possible if there is at least a world in which it is true.
The extensional analysis of language cannot account for sentences
that are very common such as those that employ opaque contexts
(to know, to believe, to think) and those that employ modal
operators (all words that can be reduced to "it is possible that"
and "it is necessary that"). These senteces are not extensional,
meaning that they do not satisfy Leibniz's law. These sentences
can be interpreted in Kripke's model-theoretic semantics. A
statement that is false in this universe can be true in another
universe. The truth values of a sentence are always relative to
a particular world.
Tarski's theory is purely extensional (for each model the truth
of a predicate is determined by the list of objects for which it
is true), Kripke's modal logic is intensional. An extensional
definition would actually be impossible, as the set of objects is
infinite.
Proper names and definite descriptions are designators. Proper
names are rigid designators, i.e. in every possible world they
designate the same object. Kripke (unlike Frege) carefully dis-
tinguishes the meaning of a designator and the way its reference
is determined (which are both "sense" in Frege). Then he puts
forth his causal theory of naming: initially, the reference of a
name is fixed by some operation (e.g., by description), then the
name is passed from link to link. A name is not identified by a
set of unique properties satisfied by the referent: the speaker
may have erronous beliefs about those properties or they may not
be unique. The name is passed to the speaker by tradition from
link to link. Terms for natural kinds behave in a similar way to
proper names.
Kripke rejects the view that either proper or common nouns are
associated with properties that serve to select their referents.
Names are just "rigid designators". Both proper and common names
have a referent, but non a Fregean sense. The property cannot
determine the reference as the object might not have that pro-
perty in all worlds. For example, gold might not be yellow in
all worlds.
Kripke's causal theory of names assumes that names are linked to
their referents through a casual chain. A term applies directly
to an object via a connection that was set in place by the ini-
tial naming of the object.
A nonrigid designator is a term that changes its referent across
possible worlds. Mental states cannot be identical to physical
states because both are rigid designators and they might desig-
nate different objects in different worlds.
Kuipers Benjamin: QUALITATIVE REASONING (MIT Press, 1994)
A unified theory of qualitative reasoning.
Qualitative reasoning is viewed as a set of methods for
representing and reasoning with incomplete knowledge about physi-
cal systems. A qualitative description of a system allows for
common sense reasoning that overcomes the limitations of rigorous
logic. Qualitative descriptions capture the essential aspects of
structure, function and behavior, at the expense of others. Since
most phenomena that matter to ordinary people depend only on
those essential aspects, qualitative descriptions are enough for
moving about in the world.
Kuipers presents his QSIM algorithm and representation for quali-
tative simulation. His model deals with partial knowledge of
quantities (through landmark values and fuzzy values) and of
change (by using discrete state graphs and qualitative differen-
tial equations). A qualitative differential equation is a quadru-
ple of variables, quantity spaces (one for each variable), con-
straints (that apply to the variables) and transitions (rules to
define the domain boundaries).
The framework prescribes a number of constraint propagation tech-
niques, including for higher-order derivatives and global dynam-
ics. First of all, it is necessary to build a model which
includes all the elements needed for simulating the system
(close-world assumption). Then the model can be simulated. The
ontological problem is solved drawing from varius techniques
(Forbus' qualitative process theory, Sussman's device modeling
approach, DeKleer's "no function in structure").
Kulas Jack, Fetzer James & Rankin Terry: PHILOSOPHY, LANGUAGE
AND ARTIFICIAL INTELLIGENCE (Kluwer, 1988)
A collection of historical articles on semantics, including
Davidson's "Truth and meaning" (1967), Grice's "Utterer's mean-
ing" (1968), Hintikka's "Semantics for propositional attitudes"
(1969), Montague's "The proper treatment of quantification in
ordinary english" (1973), Gazdar's "Phrase-structure grammar"
(1982), Stalnaker's "Possible worlds and situations".
Kulas provides a historical introduction to the field, starting
with Aristotle.
Kuppers Bernd-Olaf: INFORMATION AND THE ORIGIN OF LIFE (MIT
Press, 1990)
Kuppers thinks that all living phenomena, such as metabolism and
inheritance, can be reduced to the interaction of biological
macromolecules, i.e. to the laws of Physics and Chemistry, and,
in particular, the living cell originated from the iterative
application of the same fundamental rules that preside to all
physical and chemical processes.
The issue of the origin of life is reduced to the issue of the
origin of biological information. Information is viewed in its
different aspects: syntactic (as in information theory), semantic
(function and meaning of information for an organism's survival),
and pragmatic (following von Weiszacker, "information is only
that which produces information"). Following Manfred Eigen and
in opposition to Jacques Monod, Kuppers favors the hypothesis
that the origin of life from inorganic matter is due to emergent
processes of self-organization and evolution of macromolecules.
Natural selection applies to the molecular level.
Kuppers presents rigorous mathematical proofs of his theory,
often resorting to algorithmic theory (e.g., Gregory Chaitin's
quantitative determination of information in a structure).
Since evolution depends on the semantic aspect of information,
there is no contradiction with the second law of thermodynamics,
which is about the structural aspect of matter (i.e., the syntac-
tic aspect of information).
The origin of life is the origin of biological information. The
origin of syntactic information relates to the prebiotic syn-
thesis of biological macromolecules. The origin of semantic
information relates to the self-organization of macromolecules.
In the balance between law and chance, only the general direction
of evolution is determined by natural law: the detailed path is
mainly determined by chance. Natural law entails biological
structures, but does not specify which biological structures.
Jackendoff Ray: SEMANTIC INTERPRETATION IN GENERATIVE GRAMMAR
(MIT Press, 1972)
One of the milestone works in government and binding theory.
The author shows that theta roles determine to some extent the
wellformedness of anaphoric relations. Theta roles form a hierar-
chy and binding must respect such hierarchy by placing the
antecedent of an anaphor higher on the hierarchy than the anaphor
itself.
Jackendoff Ray: X'SYNTAX (MIT Press, 1977)(MIT Press, 1972)
A monumental study of the phrase structure of the english
language in the light of Chomsky's X-bar theory.
Jackendoff Ray: SEMANTICS AND COGNITION (MIT Press, 1983)
Jackendoff develops conceptual structures to explain language, in
a fashion similar to Fodor's mentalese.
The structure of meaning ought to be pursued on the same first
principles as phonology and syntax.
Meaning of verbs can be reduced to a few spacetime primitives,
such as motion and location.
The "extended standard theory" enhances Chomsky's standard theory
by using interpretation rules to extract the meaning of a sen-
tence. Such rules apply to the intermediate syntactic structures
used in the derivation of the phonetic representation.
Jackendoff Ray: CONSCIOUSNESS AND THE COMPUTATIONAL MIND (MIT
Press, 1987)
Jackendoff believes in a hierarchy of levels of mental represen-
tation.
The book resumes Jackendoff's claim that phonology and syntax are
key to the structure of meaning, then extends the framework
developed for language to vision and music (hinting at a possible
unification with Marr's theory of vision).
Each cognitive function exists at different levels of interpreta-
tions and cognitive functions generally interact at intermediary
levels.
Jackndoff refines and extends Fodor's idea of the modularity of
the mind.
Consciousness arises from a level of representation which is
intermediate between the sense-data and the form of thought.
Jackendoff Ray: SEMANTIC STRUCTURES (MIT Press, 1990)
Jackendoff's conceptual semantics is applied to lexical and syn-
tactic expressions in English. Jackendoff proposes a formalism
for describing lexical semantic facts and expressing semantic
generalizations. He employs multi-dimensional representations
analogous to those found in phonology.
Jackendoff Ray: LANGUAGES OF THE MIND (MIT Press, 1992)
This collection of papers summarizes Jackendoff's formal theory
on the nature of language and a modular approach to "mental ana-
tomy", and applies the same concepts to learning and common sense
reasoning.
There is a tight relationship between vision and language. A lex-
ical item contains the stereotipical image of the object or con-
cept. Knowing the meaning of a word implies knowing how the
object or concept looks like.
Jackendoff Ray: PATTERNS IN THE MIND (Harvester Wheatsheaf,
1993)
Following Chomsky, Jackendoff thinks that the human brain con-
tains innate linguistic knowledge and that the same argument can
be extended to all facets of human experience: all experience is
constructed by unconscious genetically determined principles that
operate in the brain.
The experience of spoken language is constructed by the hearer's
mental grammar: speech per se is only a meaningless sound wave,
only a hearer equipped with the proper device can make sense of
it.
These same conclusions can be applied to thought itself, i.e. to
the task of building concepts. Concepts are constructed by using
some innate, genetically determined, machinery, a sort of
"universal grammar of concepts". Language is but one aspect of a
broader characteristic of the human brain.
Jackson Frank: CONDITIONALS (Basil Blackwell, 1987)
A collection of articles by David Lewis, Robert Stalnaker, Grice
and Frank Jackson on the subject of conditionals. A theory of
conditionals must offer an account of the truth conditions of a
conditional (under which conditions "if A then B" is true or
false, or acceptable to some degree). The traditional view that a
conditional is true if and only if the antecedent is false or the
consequent is true is too simplicistic and allows conditionals
such as "if Jones lives in London, then he lives in Scotland" to
be true (if he does not live in London or lives in Scotland) when
it is obviously senseless.
Stalnaker and Lewis solve some of the problems of (subjective)
conditionals ("if it were that A then it would be that B") by
using possible-world semantics. Lewis also reviews Ernest Adams'
thesis that the assertability of (indicative) conditionals ("if A
then B") is measured by the conditional probability of the conse-
quent given the antecedent.
Jackson Frank: PERCEPTION (Cambridge University Press, 1977)
The immediate objects of perception are mental. To perceive an
object is to be in a perceptual state as a causal result of the
action of that object.
On epiphenomenal qualia Jackson proposed a famous thought
experiement: a blind neurophysiologist that knows everything of
how the brain perceives colors still cannot know what it feels
like to see a color.
Color is not a property of material things. Sense-data are not
material, they are mental.
Jauregui Jose: THE EMOTIONAL COMPUTER (Blackwell, 1995)
This is the english translation of 1990's "El Ordenador Cere-
bral".
Jauregi, like Wilson, views sociology as a branch of biology.
The same emotional system controls social, sexual and individual
behavior. Such emotional system originates from the neural
organization of the brain: emotions are rational and predictable
events. Jauregi believes that the brain is a computer, but intro-
duced the novelty of emotions as the direct product of that
computer's processing activity. It is emotions, not reason, that
directs and informs the daily actions of individuals. Jauregi
deals with humans that feel pleasure and pain rather than with
abstract problem solvers.
Jauregi begins by separating the brain and the self: the brain is
aware of what is going on in the digestive system of the body,
but will inform the self only when some correction/action is
necessary. Normally, an individual is not aware of her digestive
processes. Her brain is always informed, though. The communica-
tion channel between the brain and the self is made of emotions.
The brain can tune the importance of the message by controlling
the intensity of the emotions. Far from being an irrational pro-
cess, the emotional life is mathematically calculated to achieve
exactly the level of response needed. Feelings are subjective
and inaccessible, but they also are objective and precise.
The self has no idea of the detailed process that was going on in
the body and of the reason why that process must be corrected.
The brain's emotional system, on the other hand, is a sophisti-
cated and complex information-processing system. The brain is a
computer programmed to inform the self (through emotions) of what
must be done to preserve her body and her society. It is through
emotions that the brain informs the self of every single detail
in the body that is relevant for survival. There almost is no
instant without an emotion that tells the individual to do some-
thing rather than something else. "For human beings the reality
that ultimately matters is the reality of their feelings".
The self keeps a level of freedom: while it cannot suppress the
(emotional) messages it receives from the brain, it can disobey
them. The brain may increase the intensity of the message as the
self disobeys it a painful conflict may arise. The brain and the
self are not only separate, but they may fight each other.
Only the self can be conscious and feel, but the brain has con-
trol of both consciousness and feelings.
If we view the brain as a computer, the hardware is made of the
neural organization. There are two types of software, though:
bionatural (knowledge about the natural world) and biocultural
(such as a language or a religion). A program has three main
components: the sensory, the mental and the emotional systems.
Any sensory input can be translated automatically by the brain
into a mental (idea) or emotional (feeling) message; and vicev-
ersa. Biocultural and bionatural programs exhert emotional con-
trol over the body.
Jauregi distinguishes five systems of communication: the natural
system (the sender is a natural thing, such as a tree), the cul-
tural system (the sender is culture, something created by
humans), the somatic system (the sender is the individual's own
body), the imaginary system (the sender is imagination) and the
social system (the sender is another individual). The human
brain is genetically equipped to receive and understand all five
kinds of messages. What ultimately matters is the emotional
translations of sensory inputs.
Jaynes Julian: THE ORIGIN OF CONSCIOUSNESS IN THE BREAKDOWN OF
THE BICAMERAL MIND (Houghton Mifflin, 1977)
Jaynes makes a number of interesting points about consciousness.
Consciousness is not necessary for concepts, learning, reason or
even thinking. Awareness of an action tends to follow, not pre-
cede, the action. Awareness of an action bears little or no
influence on the outcome. Before one utters a sentence, one is
not conscious of being about to utter those specific words.
Consciousness is an operation rather than a thing. It is an
operation of analogy that transforms things of the real world
into meanings in a metaphorical space. Consciousness is a
metaphor-generated model of the world. Consciousness is based on
language, therefore it appeared after the emergence of language.
By reviewing historical documents of past civilizations, Jaynes
tries to identify when and how consciousness was born. Causes
include the advent of writing, the loss of belief in gods, epics,
and natural selection itself.
Jaynes thinks that some social institutions and religions,
psychological phenomena such as hypnosis and schizophrenia, and
artistic practices such as poetry and music are vestiges of an
earlier stage of human consciousness.
Jeanerrod Marc: THE COGNITIVE NEUROSCIENCE OF ACTION (Blackwell,
1996)
A survey of findings on the representations and processing that
lead to action, from neurophysiological data to the role of men-
tal imagery.
Johnson-Laird Philip: HUMAN AND MACHINE THINKING (Lawrence Erl-
baum, 1993)
A theory of deduction, induction and creation.
Johnson-Laird Philip: THINKING (Cambridge Univ Press, 1977)
A collection of articles that reviews the study of thinking in
the aftermath of the conceptual revolution that forced the tran-
sition from behaviorism to information-processing. Contributions
range from philosophy (Popper, Kuhn) to artificial intelligence
(Minsky, Schank).
Johnson-Laird Philip: MENTAL MODELS (Harvard Univ Press, 1983)
Johnson-Laird's representational theory assumes that mind
represents and processes models of the world. The mind solves
problems without any need to use logical reasoning. A linguistic
representation such as Fodor's is not necessary.
A sentence is a procedure to build, modify, extend a mental
model. The mental model created by a discourse exhibits a struc-
ture that corresponds directly to the structure of the world
described by the discourse.
To perform an inference on a problem the mind needs to build the
situation described by its premises. Such mental model simplifies
reality and allows the mind to find an "adequate" solution.
Johnson-Laird draws on several phenomena to prove the psychologi-
cal inadequacy of a mental logic. People often make mistakes
with deductive inference because it is not a natural way of
thinking. The natural way is to construct mental models of the
premises: a model of discourse has a structure that corresponds
directly to the structure of the state of affairs that the
discourse describes. How can children acquire inferential capa-
bilities before they have any inferential capabilities? Children
solve problems by building mental models that are more and more
complex.
Johnson-Laird admits three types of representation: "proposi-
tions" (which represent the world through sequences of symbols),
"mental models" (which are structurally analogous to the world)
and "images" (which are perceptive correlates of models).
Images are ways to approach models. They represent the perceiv-
able features of the corresponding objects in the real world.
Models, images and propositions are functionally and structurally
different.
Linguistic expressions are first transformed into propositional
representations. The semantics of the mental language then
creates correspondences between propositional representations and
mental models, i.e. propositional representations are interpreted
in mental models.
Turning to meaning and model-theoretic semantics, Johnson-Laird
proposes that a mental model is a single representative sample
from the set of models satisfying the assertion. Semantic proper-
ties of expressions are emergent properties of the truth condi-
tions. Johnson-Laird's procedural semantics assumes that there
are procedures that construct models on the basis of the meaning
of expressions.
Johnson-Laird believes that consciousness is computable. The mind
contains a high-level operating system and a hierarchy of paral-
lel processors. Conscious mind is due to a serial process of
symbolic manipulation that occurs at the higher level of the
hierarchy of processors (in the operating system), while uncons-
cious mind is due to a parallel process of distributed symbolic
representation. Emotions are non-symbolic signals, caused by
cognitive interpretations of the situation, that propagate within
the hierarchy.
Johnson-Laird Philip: THE COMPUTER AND THE MIND (Harvard Univ
Press, 1988)
An introduction to the themes and methods of cognitive science,
with a review of porduction and connectionist architectures.
Speech, vision and language are devoted long chapters. Johnson-
Laird also introduces his theory of mental models and resumes his
theory of consciousness and emotions.
Johnson-Laird Philip & Byrne Ruth: DEDUCTION (Lawrence Erlbaum,
1991)
The authors advance a comprehensive theory to explain all the
main varieties of deduction: propositional reasoning (that uses
the connectives "and", "or" and "not"), relational reasoning
(that depends on relations between entities), quantificational
reasoning (that uses quantifiers such as "any" and "some"). And
justify it with a variety of psychological experiments.
In order to understand discourse, humans construct an internal
representation of the state of affairs that is described in that
discourse. These mental models have the same structure as human
conceptions of the situations they represent. Deduction does not
depend on formal rules of inference but rather on a search for
alternative models of the premises that would refute a putative
conclusion. Central to the theory is the principle that people
use models that make explicit as little information as possible.
The theory also make sense of how people deal with conditionals.
The theory explains phenomena such as: that modus ponens ("if p
then q" and "p" then "q") is easier than modus tollens ("if p
then q" and "not q" then "not p").
Josephson John & Josephson Susan: ABDUCTIVE INFERENCE (Cambridge
University Press, 1993)
Abduction (inference to the best explanation, i.e. building the
hypothesis that best accounts for the data) is ubiquitous in
ordinary life as well as in scientific theory formation. The
book presents a dynasty of systems that explored abduction.
Intelligence is viewed as a cooperative community of knowledge-
based specialists (performing "generic tasks"). Knowledge arises
from experience by processes of abductive inference.
Jouvet Michel: LE SOMMEIL ET LE REVE (Jacob, 1992)
Jouvet was the first to localize the trigger zone for REM sleep
and dreaming in the brain stem. In this book he provides a neuro-
biological and psychological analysis of sleep and dreaming.
According to his findings, a dream is the vehicle employed by an
organism to cancel or archive the day's experiences on the basis
of a genetic program. Dreaming is a process that absorbs a lot of
energy.
This theory would also solve the dualism between hereditary and
acquired features. An hereditary component is activated daily to
decide how new data must be acquired.
Laird John, Rosenbloom Paul & Newell Allen: UNIVERSAL SUBGOALING
AND CHUNKING (Kluwer Academics, 1986)
The book describes in detail an architecture (SOAR) for general
intelligence. The universal weak method is an organizational
framework whereby knowledge determines the weak methods employed
to solve the problem, i.e. knowledge controls the behavior of
the rational agent. Universal subgoaling is a scheme whereby
goals can be created automatically to deal with the difficulties
that the rational agent encounters during problem solving.
The engine of the architecture is driven by production rules that
fire in parallel and represent task-dependent knowledge. The
architecture maintains a context which is made of four slots:
goal, problem space, state and operator. A fixed set of produc-
tion rules determines which objects have to become current, i.e.
fill those slots. In other words, they determine the strategic
choices to be made after each round of parallel processing.
A model of practice is developed based on the concept of chunk-
ing, which is meant to produce the power law of practice that
characterizes the improvements in human performance during prac-
tice at a given skill. Rosenbloom describes the XAPS3 system,
which was designed to model goal hierarchies and chunking. Each
task has a goal hierarchy. When a goal is successfully completed,
a chunk that represent the results of the task is created. In the
next instance of the goal, the system will not need to fully pro-
cess it as the chunk already contains the solution. The process
of chunking proceeds bottom-up in the goal hierarchy. The
process of chunking eventually leads to a chunk for the top-level
goal for every situation that it can encounter.
Lakoff George: METAPHORS WE LIVE BY (Chicago Univ Press, 1980)
Once metaphor is defined as the process of experiencing something
in terms of something else, metaphor turns out to be pervasive,
and not only in language but also in action and thought.
The human conceptual system is fundamentally metaphorical in
nature. Most concepts are understood in terms of other concepts.
There is a continuum that extends between subcategorization (a
category "is" another category in the sense that a category
belongs to another category) and metaphor (a category "is"
another category in the metaphorical sense).
Metaphors are used to partially structure daily concepts. They
are not random, but rather form a coherent system that allows
humans to conceptualize their experience. Metaphors create simi-
larities.
Lakoff defines three types of metaphor: "orientational" (in which
we use our experience with spatial orientation), "ontological"
(in which we use our experience with physical objects), "struc-
tural" (in which natural types are used to define other con-
cepts). Each metaphor can be reduced to a more primitive meta-
phor.
Conceptual metaphors transport properties from structures of the
physical world to non-physical structures. Language was probably
created to deal only with physical objects, and later extended to
non-physical objects by means of metaphors. The human conceptual
system is shaped by positive feedback from the environment.
Lakoff uses a theory of categories that draws from Wittgenstein's
family resemblance, Eleanor Rosch's prototype-based categoriza-
tion and Zadeh's fuzziness.
Language comprehension always consists in comprehending something
in terms of another. All our concepts are of metaphorical and are
based on our physical experience.
In accordance with Edward Sapir and Benjamin Whorf, language
reflects the conceptual system of the speaker.
Metonymy differs from metaphor in that metaphor is a way to con-
ceive something in terms of another thing, whereas metonymy is a
way to use something to stand for something else (i.e., it also
has a referential function).
Objective truth does not exist. Truth is a function of under-
standing, i.e. a function of an individual's conceptual system,
i.e. a function of coherence with such a system.
Ritual is viewed as a crucial process in preserving and propagat-
ing cultural metaphors.
Lakoff George: WOMEN, FIRE AND DANGEROUS THINGS (Univ of Chicago
Press, 1987)
Categorization is the main way that humans make sense of their
world. The traditional view that categories are defined by com-
mon properties of their members is being replaced by Rosch's
theory of prototypes. Lakoff's "experientialism" assumes that
thought is embodied (grows out of bodily experience), is imagina-
tive (capable of employing metaphor, metonymy and imagery to go
beyond the literal representation of reality), is holistic (i.e.,
is not atomistic), has an ecological structure (is more than just
symbol manipulation).
Lakoff reviews studies on categories (Wittgenstein, Berlin, Bar-
salou, Kay, Rosch, Tversky) and summarizes the state of the art:
categories are organized in a taxonomic hierarchy and categories
in the middle are the most basic. Knowledge is mainly organized
at the basic level and is organized around part-whole divisions.
Lakoff claims that linguistic categories are of the same type as
other categories.
In order to deal with categories, one needs cognitive models of
four kinds: propositional models (which specify elements, their
properties and relations among them); image-schematic models
(which specify schematic images); metaphoric models (which map
from a model in one domain to a model in another domain); and
metonymic models (which map an element of a model to another).
The structure of thought is characterized by such cognitive
models. Categories have properties that are determined by the
bodily nature of the categorizer and that may be the result of
imaginative processes (metaphor, metonymy, imagery). Thought
makes use of symbolic structures which are meaningful to begin
with. Language is characterized by symbolic models that pair
linguistic information with models in the conceptual system.
Categorization is implemented by "idealized cognitive models"
that provide the general principles on how to organize knowledge.
Lakoff George: MORE THAN COOL REASON (University of Chicago
Press, 1989)
While studying poetic metaphors, Lakoff emphasizes that metaphor
is not only a matter of words, but a matter of thought, that
metaphor is central to our understanding of the world and the
self. Poetry is simply the art of extending metaphors and there-
fore the mind's power of grasping concepts.
Langton Christopher: ARTIFICIAL LIFE (Addison-Wesley, 1989)
Proceedings of the first A-life workshop at the Santa Fe` Insti-
tute.
In Langton's own theory, living beings and cellular automata have
in common the transfer and conservation of information. Living
organisms use information, besides matter and energy, in order to
grow and reproduce. In living systems the manipulation of infor-
mation prevails over the manipulation of energy.
Life depends on a balance of information: too little information
is not enough to produce life, too much can actually be too dif-
ficult to deal with. Life is due to a reasonable amount of infor-
mation that can move and be stored. Life happens at the edge of
chaos.
Complexity is an inherent property of life. And life is a pro-
perty of the organization of matter.
In order to build artificial life Langton defines a "generalized
genotype" as the set of low-level rules serving as the genetic
blueprint and the "generalized phenotype" as the structure that
is created from those instructions.
Langton Christopher: ARTIFICIAL LIFE II (Addison-Wesley, 1992)
Proceedings of the second A-life workshop at the Santa Fe` Insti-
tute.
Langton Christopher: ARTIFICIAL LIFE (MIT Press, 1995)
A collection of articles by various authors originally published
in the Artificial Life journal.
Larson Richard & Segal Gabriel: KNOWLEGDE OF MEANING (MIT Press,
1995)
An introduction to truth-theoretic semantics for natural
languages, viewed as part of cognitive psychology. Unlike most
semantic studies, which are based on Montague's semantics, this
one is from Davidson's perspective.
Lashley Karl Spencer: BRAIN MECHANISMS AND INTELLIGENCE (Dover,
1963)
This 1929 study set the standard for cognitive neurophysiology
and psychology. The 1963 reissue comes with a preface by Donald
Hebb that puts Lashley's achievements in perspective.
In Lashley's mnemonic distribution model each mnemonic function
is not localized in a specific point of the mind, but distributed
over the entire mind.
Later Lashley also noted how the dualism between mind and brain
resembles the one between waves and particles. A memory in the
brain behaves like a wave in an electromagnetic field.
Laszlo Ervin: INTRODUCTION TO SYSTEMS PHILOSOPHY (Gordon &
Breach, 1972)
Von Bertalanffy's general systems theory lends itself to a
natural wedding of scientific information and philosophic mean-
ing. General systems theory consists in the exploration of
"wholes", which are characterized by such holistic properties as
hierarchy, stability, teleology. Laszlo advocates a return from
analytic to synthetic philosophy.
Laszlo starts by offering his own take at a "theory of natural
systems" (i.e., a theory of the invariants of organized complex-
ity). At the center of his theory is the concept of "ordered
whole" (a non-summative system subject to a set of constraints
that define its structure and allow it to achieve adaptive self-
stabilization). Laszlo then adopts a variant of Ashby's principle
of self-organization, according to which any isolated natural
system subject to constant forces is inevitably inhabited by
"organisms" that tend towards stationary or quasi-stationary
non-equilibrium states. In Laszlo's view the combination of
internal constraints and external forces yields adaptive self-
organization. Natural systems evolve towards increasingly adapted
states, corresponding to increasing complexity (or negative
entropy).
Natural systems sharing an environment tend to organize in
hierarchies. The set of such systems tends to become itself a
system, its subsystems providing the constraints for the new sys-
tem.
Laszlo then offers rigorous foundations to deal with the emer-
gence of order at the atomic ("micro-cybernetics"), organismic
("bio-cybernetics") and social levels ("socio-cybernetics").
A systemic view also permits a formal analysis of a particular
class of natural systems, that of cognitive systems. The mind,
just like any other natural system, exhibits an holistic charac-
ter, adaptive self-organization, and hierarchies, and can be stu-
died with the same tools ("psycho-cybernetics").
The basic building blocks of reality are therefore natural sys-
tems.
Lavine Robert: NEUROPHYSIOLOGY (Collamore, 1983)
A comprehensive introduction to the neuron, the structure of the
brain, senses and to higher cognitive functions.
Layzer David: COSMOGENESIS (Oxford University Press, 1990)
Inspired by cosmology, Layzer deals with the paradox of creation
of order by saying that, if entropy in the environment increases
more than the entropy of the system, then the system becomes more
ordered in that environment. Entropy and order can both increase
at the same time without violating the second law of thermodynam-
ics. This phenomenon can be described as: if the expansion of a
set of systems is so quick that a number of states which are
occupied increases less rapidly than the number of states which
are available (i.e., the phase space gets bigger), entropy and
order can increase at the same time.
Unlike Prigogine, Layzer does not need to assume that an energy
flow from the environment of a system can cause a local decrease
in entropy within the system. Entropy and order increase
together because the realization of structure lags behind the
expansion of phase space.
Drawing from Shannon's theory of communication, David Layzer
defines information as the difference between potential entropy
(the largest possible value that the entropy can assume under the
specified conditions) and actual entropy. As actual information
increases, actual entropy decreases (information is "negative"
entropy in Shannon's theory). Potential entropy is also poten-
tial information: maximum entropy equals maximum information.
In biological and astronomical systems the potential entropy may
increase with time, thereby creating information if it increases
faster than actual entropy. In particular, both contraction and
expansion of the universe from an initial state of thermodynamic
equilibrium would generate potential entropy. Genetic variation
always generates entropy as information flows unidirectionally
from the genotype to the phenotype: when it makes the distribu-
tion of genotypes more uniform in a genotype space, it generates
entropy and destroys information; when it allows the population
to populate previously uninhabited regions fo the genotype space,
it generates potential entropy without necessarily generating
entropy.
Layzer then proves that several evolutionary processes (mutation,
differential reproduction, gene duplication, differentiation and
integration) generate biological information. Natural itself
selection always increases the proportion of relatively fit vari-
ants in a population and decreases the proportion of relatively
unfit variants, therefore natural selection always generates bio-
logical order.
Layzer thinks that biological evolution is not driven by the
growth of entropy (as a counterweight to the loss of order), it
is not (directly or indirectly) driven by the second law of ther-
modynamics. That law presupposes certain initial and boundary
conditions that are not present in biological systems.
Influenced by Schmalhausen's theory that evolution is a process
of hierarchical construction, Layzer thinks that there is a sin-
gle universal law governing processes that dissipate order, but
order is also generated by several hierarchically linked
processes (including cosmic expansion and biological evolution).
Lazarus Richard: EMOTION AND ADAPTATION (Oxford Univ Press,
1991)
Lazarus argues that the final goal of our emotions is to help the
organism survive in the environment. His theory is a "relational"
theory of emotions, in that it assumes that emotions arise from
an adaptational situation of the individual in the environment.
Emotions are reactions to attempted goals ("motivational princi-
ple"), emotions represent reactions to evaluations of relation-
ships with the environment. Stable relationships between the
individual and the environment result in recurrent emotional pat-
terns in the individual. Emotion is due to an evaluation of the
potential consequences of a situation.
The development of the self is a fundamental event for the emo-
tional life. Emotions depend on an organizing principle in which
the self is distinguished from the non-self, because only after
that principle has become established can evaluations of benefits
and harms be performed. Differentiation of self and other is a
fundamental property of living organisms (even plants use protein
discrimination mechanisms, and most organisms could not survive
without the ability to distinguish alien organisms).
An emotion is a process in four stages: anticipation, provoca-
tion, unfolding, outcome. Both biological and social variables
contribute to this process, and this explains why emotions change
through the various stages of life.
Each type of emotion can be defined by a relational meaning which
expresses the set of benefits and harms in a relationship between
individual and environment and is constructed by a process of
appraisal. Each type of emotion is distinguished by a pattern of
appraisal factors. The relational meaning is about the signifi-
cance of the event for the well-being of the individual. Emo-
tions express the personal meaning of an individual's experience.
Lazarus, unlike Zajonc, emphasizes cognition in the relationship
between emotion and cognition. After all, appraisal is the fun-
damental process for the occurrence of emotion.
Lazarus Richard & Lazarus Bernice: PASSION AND REASON (Oxford
Univ Press, 1994)
Lazarus reiterates his point that emotions are as rational as
anything can be in a language accessible to anybody.
Ledoux Joseph & William Hirst: MIND AND BRAIN (Cambridge Univ
Press, 1986)
A collection of articles on perception, attention, memory and
emotion that are organized as debates between psychologists and
neurobiologists.
Lehnert Wendy: STRATEGIES FOR NATURAL LANGUAGE LANGUAGE
(Lawrence Erlbaum, 1982)
A practical textbook on natural language processing in the con-
ceptual dependency tradition. Each chapter is written by an
authority of the field. Includes Steven Small's word-based
parser, Gerald DeJong's FRUMP system, Wilensky's PAM system,
Wendy Lehnert's plot units, Schank's MOPs. Jerry Hobbs writes
about coherence in discourse. Yorick Wilks discusses procedural
semantics.
Leiser David & Gillieron Christiane: COGNITIVE SCIENCE AND
GENETIC EPISTEMOLOGY (Plenum Press, 1989)
The book analyzes the relations between procedures and structures
from a Piagetian perspective and attempts to bridge a gap between
cognitive psychology and artificial intelligence.
Lenat Douglas: BUILDING LARGE KNOWLEDGE-BASED SYSTEMS (Addison-
Wesley, 1990)
The book describes the CYC system, whose goal is to represent
common knowledge (i.e., develop a global ontology) and perform
common-sense reasoning (i.e., employ a set of reasoning methods
as a set of first principles) on large knowledge bases. to
explain
Units of knowledge for common sense are units of "reality by con-
sensus": all the things we know and we assume everybody knows;
i.e., all that is implicit in our acts of communication. A prin-
ciple of economy of communications states the need to minimize
the acts of communication and maximize the information that is
transmitted. World regularities belong to this tacitly accepted
knowledge.
Lenneberg Eric: BIOLOGICAL FOUNDATIONS OF LANGUAGE (Wiley, 1967)
Language should be studied as an aspect of man's biological
nature, in the same manner as anatomy. Chomsky's universal gram-
mar is to be viewed as an underlying biological framework for the
growth of language. Genetic predisposition, growth and develop-
ment apply to language faculties just like to any other organ of
the body. Behavior in general is an integral part of an
organism's constitution.
Language and speech are represented in the cortex and also seem
to be hosted in subcortical and midbrain structures. The large
size of the human brain is probably a direct consequence of
language functions. Children start learning language when struc-
tural changes in the brain make it possible.
Animals organize the sensory world through a process of categori-
zation. They exhibit propensities for responding to categories
of stimuli. In humans this process of categorization becomes
"naming", the ability to assign a name to a category. Even in
humans the process of categorization is still a process whose
function is to enable similar response to different stimuli. The
meaning-bearing elements of language do not stand for specific
objects, but for the act of categorization. The basic cognitive
mechanisms of semantics are processes of categorization
LePore Ernest: NEW DIRECTIONS IN SEMANTICS (Academic Press,
1987)
A collection of articles on semantics, including Hintikka's
game-theoretical semantics, Gilbert Harman's conceptual role
semantics (the ultimate source of meaning is the functional role
that symbols play in thought) and dual aspect semantics (which
contain one theory relating language to the world and one theory
relating language to the mind).
Lesniewski Stanislaw: COLLECTED WORKS (Kluwer Academic, 1991)
In the Thirties the polish logician Lesniewski noted that in any
language containing its semantics logical laws cannot hold con-
sistently. A contradiction can be avoided only by reconstructing
the object language through hierarchical levels, or
metalanguages. This is similar to Russell's conclusion that some
hierarchy is necessary for a system to be coherent. Lesniewski
developed a hierarchy of categories (a grammar of semantic
categories). Lesniewski's system consists of three axiomatic
theories: protothetic (a calculus of equivalent propositional
functions, with a single axiom), ontology (a calculus of classes
in terms of a theory of nominal predication, with a single axiom)
and mereology (based on the part-whole relation, containing rules
to avoid paradoxes). Functorial categories can be generated from
a set of basic categories (the propositions defined by the single
axiom of protothetic and the nouns defined by the single axiom of
ontology) and are categories of functions from certain arguments
to certain values.
Levine Daniel: INTRODUCTION TO NEURAL AND COGNITIVE MODELING
(Lawrence Erlbaum, 1991)
A broad survey of cognitive science from a neuroscientific per-
spective. After a historical outline (McCulloch-Pitts neurons,
Hebb's law, Rosenblatt's perceptron, etc), Levine details algo-
rithms (and physiological justifications) for associative learn-
ing, competition, conditioning, categorization, representation.
All the main connectionist models are surveyed. The book pro-
vides a detailed, technical compendium of data and ideas in the
field.
Levinson Stephen: PRAGMATICS (Cambridge Univ Press, 1983)
An excellent and relatively accessible introduction to pragmat-
ics.
Levinson surveys the issues of pragmatics, defined essentially as
the relationship between language and context. Approaches to
indexicals or deixis (Fillmore, Lyons, Lakoff), implicatures
(Grice, Gazdar), presupposition (Stalnaker, Karttunen), speech
acts (Austin, Searle), and discourse analysis are dealt with at
length. This is the best introduction to the theories that
emerged during the late Seventies.
Levy Steven: ARTIFICIAL LIFE (Pantheon, 1992)
An introduction for the wider audience to the world of artificial
life. Includes history of the field (from Von Neumann to
viruses), biographies of its visionaries (Kauffman, Holland, Haw-
kins, Ray, Brooks) and simplified presentations of their
theories.
Lewin Roger: COMPLEXITY (Macmillan, 1992)
Complexity is presented as a discipline that can unify the laws
of physical, chemical, biological, social and economic phenomena
through the simple principle that all things in nature are driven
to organize themselves into patterns. The book, written in
conversational english, devotes much time to describing the pro-
tagonists of the field and relating interviews in a celebrity-
centered fashion.
Lewis Clarence Irving: SYMBOLIC LOGIC (Mineola, 1932)
In Lewis' modal logic a proposition is necessary if it is true in
every possible world, it is possible if it is true in at least
one possible world. "Necessity" and "possibility" are modal
operators, i.e. they operate on logical expressions just like
logical connectives. The two modal operators are dual (one can
be expressed in terms of the other), thereby reflecting the dual-
ism of the two corresponding quantificators (existential and
universal). A modal logic is built by adding a few axioms con-
taining the modal operators to the axioms of a non-modal logic.
Lewis David K.: COUNTERFACTUALS (Harvard Univ Press, 1973)
Lewis uses possible-world semantics in his theory of counterfac-
tuals. Lewis defines a pair of conditional operators ("if it
were the case that, then it would be the case that" and "if it
were the case that, then it might be the case that"), which can
be defined one in terms of the other. Counterfactuals are not
strict conditionals (material conditionals preceded by a neces-
sity operator), but rather "variably" strict conditionals (a
counterfactual is as strict as it must be to escape vacuity and
no stricter).
Lewis then defends possible worlds and claims that each possible
world is as "real" as ours.
Lewis also compares his theory to Stalnaker's own, which is also
based on possible worlds.
Lewis David K.: PHILOSOPHICAL PAPERS (Oxford Press, 1983)
The ultimate collection of Lewis' papers.
In "An argument for the identity theory" Lewis argues that a men-
tal state can be defined by a physical state, which is not neces-
sarily the same for all species, and by a "causal role", that
expresses behavior that such a state induces in the organism.
In "Radical interpretation" he contends that intentional ascrip-
tion (the task of redescribing the information of an individual
in intentional terms) is a kind of constraint-satisfaction prob-
lem: the correct intentional ascription is the one that provides
a best fit to the demands that the constraints impose. The prob-
lem of radical interpretation is tackled by identifying four
parts: the intentional system (e.g., a person); the system's
attitudes (beliefs and desires) as expressed in the observer's
language; the system's attitudes (beliefs and desires) as
expressed in the system's own language; and the system's mean-
ings.
The constraints are derived from six principles: the principle of
charity constrains the relation between a system and its beliefs
expressed in the observer's language (the system's beliefs are
somehow constrained by the observer's beliefs); the rationaliza-
tion principle constrains the relation between the beliefs
expressed in the observer's language and the system (the system
is a rational agent, his beliefs being what make sense for its
behavior); the principle of truthfulness constrains the relation
between the observer's beliefs and the system's meanings; the
principle of generativity constrains the meaning in that it
should assign truth conditions to the system's sentences in a
reasonable way; the manifestation principle constrains the rela-
tion between the system and its beliefs (they must be consistent
with its speech behavior); and the triangle principle constrains
the relation between the system's meaning, its beliefs and the
observer's beliefs.
Lewis David K.: ON THE PLURALITY OF WORLDS (Basil Blackwell,
1986)
Lewis advocates an indexical theory of actuality. Every possible
world is actual from its own point of view, and every possible
world is merely possible from the point of view of other worlds.
Worlds are never causally related to other worlds. The isolation
of possible worlds constitutes their being merely possible rela-
tive to each other.
A proposition is a function from possible worlds to truth-values.
Each world provides a truth value for a proposition.
Lewontin Richard: THE GENETIC BASIS OF EVOLUTIONARY CHANGE
(Columbia University Press, 1974)
It is not yet clear which percentage of evolutionary change is
due to natural selection and which is due to random events.
Modern evolutionary genetics stems from the merging of two tradi-
tions, the Darwinian and the Mendelian, both of which take varia-
tion as the crucial aspect of life. The Darwinian view can be
summarized as "evolution is the conversion of variation between
individuals into variation between populations and species in
time and space". The paradox is that Mendelian theory dictates
the frequencies of genotypes as the appropriate genetic descrip-
tion of a population, whereas variation is much more important.
"What we can measure is uninteresting and what we are interested
in is unmeasurable". Most theories of genetic variation in popu-
lations (allelic variation) are also theories of natural selec-
tion. Variation and selection turn out to be dual aspects of the
same problem.
Even worse is the situation with respect to "the origin of
species", i.e. theories of the genetic changes that occur in
species formation. Geographic isolation (or, better, ecological
divergence) is recognized as the preliminary stage, causing the
appearance of genetic differences sufficient to restrict severely
the genetic exchange with other populations (reproductive isola-
tion). The second stage occurs when isolated populations come
into contact and the third stage starts when the newly formed
species continue to develop independently.
Lewontin reviews evidence in favor of each theory. His conclu-
sion, in ragarding the genome as the uit of selection, is that
"context and interaction are of essence".
Lewontin Richard: HUMAN DIVERSITY (W.H.Freeman, 1981)
Each organism is the subject of continous development throughout
its life and such development is driven by mutually interacting
genes and environment. Genes per se cannot determine the pheno-
type, capacity or tendencies.
The organism is both the subject and the object of evolution.
Organisms construct environments that are the conditions for
their own further evolution and for the evolutions of nature
itself towards new environments. Organism and environment mutu-
ally specifify each other.
Leyton Michael: SYMMETRY, CAUSALITY, MIND (MIT Press, 1992)
Leyton's idea is that shape is used by the mind to recover the
past. Shape is time. Shape equals the history that created it.
By studying the psychological relationship between shape and
time, Leyton offers a working model of how perception is con-
verted into memory.
There is a relationship between perceived asymmetry, inferred
history and environmental energy. The energy of a system
corresponds to memory of the causal interactions that transferred
to the system.Shape, or asymmetry, is a memory of the energy
transferred to an object in causal interactions.
All vision is the recovery of the past: vision simply "unlocks"
time from the image. In general, perceptual representations are
representations of stimuli in terms of causal histories. This is
also true of cognitive representations.
Any cognitive representation is the description of a stimulus as
a state in a history that causally explains the stimulus to the
organism. A cognitive system is a system that creates and mani-
pulates causal explanations.
Ming Li and Paul Vitanyi: AN INTRODUCTION TO KOLMOGOROV COMPLEX-
ITY AND ITS APPLICATIONS (Springer-Verlag, New York, 1997)
Written by two experts in the field, this is the only comprehen-
sive and unified treatment of the central ideas and their appli-
cations of Kolmogorov complexity (the theory dealing with the
quantity of information in individual objects), also known vari-
ously as `algorithmic information', `algorithmic entropy',
`Kolmogorov- Chaitin complexity', `descriptional complexity',
`shortest program length', `algorithmic randomness', and others.
Lieberman Phipip: THE BIOLOGY AND EVOLUTION OF LANGUAGE (Harvard
Univ Press, 1984)
Language is found to be a by-product of the neural processes that
underly cognition in general (unlike Chomsky's vision of separate
"language organs"). The only language-specific processes are
essentially those that contribute to speech, and they evolved
from processes that are common to many animals. Speech, not syn-
tax, is the fulcrum of language.
Lieberman Philip: UNIQUELY HUMAN (Harvard Univ Press, 1992)
Human language is a relatively recent evolutionary innovation
that came about when speech and syntax were added to older com-
munication systems. The function of speech and syntax is to
enhance the speed of communication: speech allows humans to over-
come the limitations of the mammalian auditory system and syntax
allows them to overcome the limits of memory.
Two principles are recalled. Natural selection acts on individu-
als who each vary: species that successfully change and adapt are
able to maintain a stock of varied traits coded in the genes of
the individuals who make up their population. The "mosaic" prin-
ciple states that parts of the body of an organism are governed
by independent genes. There are no central genes who control the
overall assembly of the body.
Given these principles, two phenomena can be explained: a series
of small, gradual changes in structure can lead to an abrupt
change in the behavior of the organism; and an abrupt change in
behavior may cause an abrupt change in morphology which causes
the formation of a new species (at "functional branch-points").
The structure of the brain reflects its evolutionary history. The
brain consists of a set of specialized circuits with independent
evolutionary histories. Unlike modular theories such as
Chomsky's and Fodor's, Lieberman's "circuit model" (derived from
Norman Geschwind's connectionist model) assumes that the brain
bases for language are mostly language-specific and mostly
located in the newest part of the brain, the neocortex. The
brain consists of many specialized units that work together in
different circuits (the same unit can work in many circuits). The
overall circuitry reflects the evolutionary history of the brain,
with units that adapted to serve a different purpose from their
original one. Therefore, rapid vocal communication is responsible
for the evolution of the human brain.
The theory is supported by a wealth of anthropological and neuro-
physiological data (particularly from Broca's and Wernicke's
experiments).
Besides language, the other unique trait of the human race (and
therefore of the human brain) is moral code, in particular
altruism. This is also a relatively recent development, and
presupposes language and cognition.
Lightfoot David: THE LANGUAGE LOTTERY (MIT Press, 1982)
The book is basically an introduction to Chomsky's theories of
language with an emphasis on biological aspects. First and
foremost, Lightfoot examines how children can learn a language
without significant instruction and despite a deficiency of
experiential data. The only rational explanation is that an
innate structure, a "universal grammar", guides the learning pro-
cess.
Lightfoot applies Gould's theory of evolutionary change to
linguistics: language changes gradually but every now and then is
subject to catastrophic revisions.
Lockwood Michael: MIND, BRAIN AND THE QUANTUM (Basil Blackwell,
1989)
Drawing from quantum mechanics and from Bertrand Russell's idea
that consciousness provides a kind of "window" onto the brain,
Lockwood offers a theory of consciousness as a process of percep-
tion of brain states.
By using special relativity (mental states must be in space given
that they are in time), he leans towards the identity theory.
Then Lockwood interprets the role of the observer in quantum
mechanics as the role of consciousness in the physical world (as
opposed as a simple interference with the system being observed).
Lockwood thinks that our sensations are intrinsic attributes of
physical states of the brain. Consciousness scans the brain to
look for sensations. It does not create them, it just seeks them.
Each observable attribute (e.g., each sensation) corresponds to
an observable of the brain. The external world is a physical sys-
tem in which a set of compatible observables is defined, whose
state is therefore defined by a sum of eigenstates of such
observables (i.e., by a sum of perspectives).
Lockwood mentions Deutsch David's "Quantum theory and the univer-
sal quantum computer" (1975), which generalizes Turing's ideas
and defines a "quantum" machine in which Turing states can be
linear combinations of states. The behavior of a quantum machine
is a linear combination of the behavior of several Turing
machines. A quantum machine can only compute recursive functions,
as much as Turing's machine, but it turns out to be much faster
in solving problems that exhibit some level of parallelism. In a
sense a quantum computer is capable of decomposing a problem and
delegating the subproblems to copies of itself in other
universes.
Luger George: COMPUTATION AND INTELLIGENCE (MIT Press, 1995)
Seminal papers by Turing, Minsky, McCarthy, Newell, Schank,
Brooks.
Luger George: COGNITIVE SCIENCE (Academic Press, 1993)
An introduction to the field.
Lukaszewicz Witold: NON-MONOTONIC REASONING (Ellis Harwood,
1990)
A formal survey of mathematical theories for nonmonotonic reason-
ing.
After an introduction to monotonic logic (first and second
order), the book delves into nonmonotonic logics: Sussman's
MICRO-PLANNER, Doyle's and de Kleer's truth maintenance systems,
Mc Carthy's circumscription, McDermott and Doyle's modal logic,
Moore's autoepistemic logic, Reiter's default logic and the
closed world assumption.
Lycan William: LOGICAL FORM IN NATURAL LANGUAGE (MIT Press,
1984)
Lycan's theory of linguistic meaning rests on truth conditions.
All other aspects of semantics (verification conditions, use in
language games, illocutionary force, etc) are derived from that
notion. A sentence is meaningful in virtue of being true under
certain conditions and not others.
This is consistent with Davidson's program of assigning meanings
to sentences of natural languages by associating the sentences
with truth-theoretically interpreted formulas of a logical system
(their "logical form"). Lycan basically refines Davidson's
metatheory. Instead of assigning only a pair of arguments to the
truth predicate, Lycan defines truth as a pentadic relation with
reading (the logical form), context (truth is relative to a con-
text of time and speaker, as specified by some assignment func-
tions), degree (languages are inherently vague, and sentences
normally contain fuzzy terms and hedges), and idiolect (the truth
of a sentence is relative to the language of which it is a gram-
matical string).
Lycan argues that pragmatics (implicatures, presuppositions)
should be kept separate from semantics. Context determines the
interpretation of a sentence at several levels: it singles a log-
ical form out of a set of potential candidates; it completes its
proposition by binding all free variables; it provides a secon-
dary meaning (e.g., implicatures); it clarifies lexical presump-
tions; and it determines the illocutionary force.
Lycan defends truth-condition semantics against the most common
attacks, in particular against Quine's theory of indeterminacy
and Dummett's antirealism.
Lycan finally presents a cognitive architecture based on a ver-
sion of humuncular functionalism.
Lycan William: CONSCIOUSNESS (MIT Press, 1987)
Lycan reviews behaviorist and dualist theories of the mind, then
focuses on Dennett's homuncular functionalism and defends it
against its critics.
Lycan thinks that, besides the low level of physiochemical
processes and the high level of psychofunctional processes,
Nature is organized in a number of hierarchical levels (suba-
tomic, atomic, molecular, cellular, biological, psychological).
And each level is both physical and functional: physical with
respect to its immediately higher level and functional with
respect to its immediately lower level. Going from lower levels
to higher levels we obtain a physical, structural, description of
nature (atoms make molecules that make cellules that make organs
that make bodies...). Backwards we obtain a functional descrip-
tion (the behavior of something is explained by the behavior of
its parts).
The aggregative ontology ("bottom-up") and the structured
epistemology ("top-down") of Nature are dual aspects of the same
thing. The apparent irreducibility of the mental is due to the
irreducibility of the various levels.
Lycan William: MIND AND COGNITION (MIT Press, 1990)
A massive collection of articles on theories of the mind. Homun-
cular functionalism is championed by Dennett and Lycan. Elimina-
tivism is presented by Churchland and Feyerabend. Language of
thought (Fodor), folk psychology (Stich), qualia (Block) are also
discussed.
Lycan William: MODALITY AND MEANING (Kluwer Academic, 1994)
Lycan presents a theory of possible individuals and possible
worlds in which a world is viewed as a structured set of proper-
ties. A number of philosophical puzzles are examined from a very
technical perspective.
Lycan William: CONSCIOUSNESS AND EXPERIENCE (MIT Press, 1996)
A continuation of "Consciousness".
Lyons John: SEMANTICS (Cambridge Univ Press, 1977)
A discussion of semantics within the framework of semiotics, i.e.
taking language as a semiotic system.
Lyons discusses behaviorist semantics, logical semantics (model-
theoretic and truth-conditional semantics, reference, sense and
naming) and structuralist semantics (in particular semantic
fields and componential analysis).
The second volume is more specifically linguistic, dealing with
grammar, deixis, illocutionary force, modality.
MacCormac Earl: A COGNITIVE THEORY OF METAPHOR (MIT Press, 1985)
A unified theory of metaphor, with implications for meaning and
truth.
MacCormac rejects the tension theory (which locates the differ-
ence between metaphor and analogy in the emotional tension gen-
erated by the juxtaposition of anomalous referents), Monroe
Beardsley's controversion theory (which locates that difference
in the falsity produced by a literal reading of the identifica-
tion of the two referents) and the deviance theory (which locates
that difference in the ungrammaticality of the juxtaposition of
two referents). A metaphor is recognized as a metaphor on the
basis of the semantic anomaly produced by the juxtaposition of
referents. Metaphor is distinct from ordinary language (as
opposed to the view that all language is metaphorical).
MacCormac modifies Black's interactionist theory and adopts
Wheelwright's classification of "epiphors" (metaphors that
express the existence of something) and "diaphors" (metaphors
that imply the possibility of something). Diaphor and epiphor
measure the likeness and the dissimilarity of attribute of the
referents. A diaphor can become an epiphor (when the object is
found to really exist) and an epiphor can become a literal
expression (when the term has been used for so long that people
have forgotten its origin).
Metaphor is a process that exists at three levels: a language
process (from ordinary language to diaphor to epiphor back to
ordinary language); a semantic and syntactic process (its
linguistic explanation); and a cognitive process (to acquire new
knowledge). Therefore a theory of metaphor requires three lev-
els: a surface or literal level, a semantic level and a cogni-
tive level.
The semantics of metaphor is then formalized using mathematical
tools. "Partial" truths of metaphorical language are represented
by fuzzy values: the meaning of a sentence can belong to several
concepts with different degrees of memberships. The paradigm is
one of language as a hierarchical network in n-dimensional space
with each of the nodes of the network a fuzzy set (defining a
semantic marker). When unlikely markers are juxtaposed, the
degrees of membership of one semantic marker in the fuzzy set
representing the other semantic marker can be expressed in a
four-valued logic (so that a metaphor is not only true or false).
MacCormac also sketches the theory that metaphors are speech acts
in Austin's sense. Metaphors both possess meaning and carry out
actions. An account of their meaning must include an account of
their locutionary and perlocutionary forces.
Finally, the third component of a theory of meaning for metaphors
(besides the semantic and speech act components) is the cultural
context.
The meaning of metaphors results from the semantical aspects of
communication, culture and cognition.
MacCormac claims that, as cognitive processes, metaphors mediate
between culture and the mind, influencing both cultural and bio-
logical evolution.
MacLean Paul: THE TRIUNE BRAIN IN EVOLUTION (Plenum Press, 1990)
From the study of stimulation and lesion of different brain
areas, MacLean developed his theory of the "triune brain": the
human brain is divided in three layers, that correspond to three
different stages of evolution. The oldest one, the "reptilian
brain", is a midbrain reticular formation that has changed little
from reptiles to mammals and to humans and is responsible for
species-specific behavior (instinctic behavior). The limbic sys-
tem is the old mammalian brain which is responsible for emotions
that are functional to the survival of the individual and of the
species. The cerebral cortex is the new mammalian brain, which is
responsible for higher cognitive functions such as language and
reasoning.
MacNamara John & Reyes Gonzalo: THE LOGICAL FOUNDATIONS OF COG-
NITION (Oxford University Press, 1994)
A collection of papers that try to bridge logic and cognition.
The editors believe that the most basic properties of cognitive
psychology show up as the universal properties of category
theory. Category theory is better suited than set theory for
representing basic intentional capabilities such as to refer, to
count and to learn. Category theory generalizes set theory's
notions of set and function into the "universal" properties of
object and morphism. Logic becomes the study of what is univer-
sal.
The editors reach the conclusion that "there is no purely physio-
logical explanation for the acquisition of intentional skills or
the existence of intentional states." As a corollary, there must
exist unlearned (innate) "logical resources" (e.g., membership,
typed equality, reference to symbols), sort of universals of the
human mind.
Most papers revolve around Reyes' seminal contribution to a
semantic theory. Kinds are interpretations of common nouns.
Reference to an individual by means of a proper noun involves a
kind (e.g., reference to the name of a person involves the kind
"person"). Therefore any reference to an individual involves a
kind. Kinds are modally constant (don't decay in time), but
predicates (properties) of kinds may change. All predicates are
typed by kinds.
Maes Patti: DESIGNING AUTONOMOUS AGENTS (MIT Press, 1990)
A collection of articles on action-oriented systems (as opposed
to knowledge-based systems), which are based on a tighter cou-
pling between perception and action, a distributed architecture,
dynamic interaction with the environment. Cognitive faculties are
viewed as "emergent functionalities", properties that arise from
the interaction of the system with the environment. The proper-
ties of the environment determine the behavior of the system as
much as the system's own properties. A system is made of auto-
nomous specialized subsystems and the overall behavior is the
result of the intended behaviors of all the subsystems.
Rodney Brooks introduces his situated agents.
Mae models action selection through behavior networks, which
exhibit planning capabilities halfway between traditional goal-
oriented planning and situated action.
Mamdani E.H. & Gaines B.R.: FUZZY REASONING (Academic Press,
1981)
Each chapter is written by an authority in the field. Zadeh
introduces PRUF, a meaning representation language for natural
languages that considers the intrinsic imprecision of languages
as possibilistic rather than probabilistic in nature. Goguen pro-
vides some mathematical foundations to the theory of fuzzy sets,
that lead from a few axioms (in the language of category theory)
to the definition of operations on fuzzy sets that are parallel
to those for ordinary sets (which, on the other hand, cannot be
categorical).
Some applications to linguistics, expert systems and controllers
are also discussed.
Mandelbrot Benoit: THE FRACTAL GEOMETRY OF NATURE (W.H.Freeman,
1982)
This is the book (a revision of 1977's "Fractals") that made
fractals popular. Mandelbrot emphasizes the inhability of clas-
sical geometry to model the shapes of the real world, in particu-
lar the complexity of natural patterns. Natural patterns (such
as coastlines) are of infinite length. In order to provide meas-
ures, Mandelbrot resorts to Felix Hausdorff's fractal dimension
(a fraction that exceeds one, even if a curve's dimension should
intuitively be one).
Scaling and nonscaling fractals, self-mapping fractals, Brown
fractals, trema fractals are introduced along with their
mathematical properties.
Mandelbrots describes applications to coastal lines, galaxy clus-
ters, the physics of turbulence, the cosmological principle, and
so forth, and discusses the relationship to artificial life
(organic looking nonlinear fractals) and chaos theory (nonlinear
fractals that play the role of attractors for dynamic systems).
Mandler George: MIND AND BODY (Norton, 1984)
An expanded and revised edition of "Mind and Emotion" (1975),
which first analyzed the relationship between cognition and emo-
tion.
After a generous history and survey of research on emotions in
cognitive psychology, Mandler offers his view on mind and cons-
ciousness: the mind is a general information-processing system
that employs schemas as basic cognitive structures. Schemas
represent environmental regularities.
Mandler emphasizes the constructive nature of consciousness:
"consciousness is a construction of phenomenal experience out of
one or more of the available preconscious schemas," a process
driven by the most abstract schema relevant to the current goals
of the individual. One of the functions of consciousness is to
enable the individual to evaluate environmental conditions and
action alternatives.
Emotions are constructed out of autonomic arousal (arousal of a
part of the nervous system called autonomic nervous system, which
determines the intensity of the emotion) and evaluative cognition
(meaning analysis, which determines the quality of the emotion).
Therefore, emotion is a product of schemas, arousal and cons-
ciousness. The function of emotions is to provide the individual
with an optimal sense of the world, with the most general picture
of the world that is consistent with current needs, goals and
situations.
Marcus Mitchell: A THEORY OF SYNTACTIC RECOGNITION FOR NATURAL
LANGUAGE (MIT Press, 1980)
This book describes the famous Marcus parser.
Marek Wiktor & Truszczynski Miroslav: NON-MONOTONIC LOGIC
(Springer Verlag, 1991)
A rigorous, monumental work on the foundations of nonmonotonic
logic, based on nonmonotonic rules of proof (defaults), or
context-dependent derivation (the context determines which
derivation rule is valid). First-order default theories such as
Reiter's and modal nonmonotonic logics such as Doyle's and
McDermott's are given extensive treatments, while second-order
logics such as McCarthy's circumscription are merely mentioned.
Margalef Ramon: PERSPECTIVES IN ECOLOGICAL THEORY (Univ of Chi-
cago Press, 1968)
In this study of the ecosystem as a cybernetic system a number of
biological quantities are given mathematical definition.
A basic property of nature is that any exchange between two sys-
tems of information increases the difference of information
between the two systems: the less organized system gives energy
to the more organized one and in parallel information is des-
troyed in the less organized system and information is created in
the more organized one. The less organized system feeds the more
organized.
An ecosystem is a system controlled by the second law of thermo-
dynamics.
A measure of ecological efficiency is given by the energy flow
per unit biomass (the primary production of the system divided
by the total biomass).
Succession (the occupation of a territory by organisms) is a
self-organizing process that develops a biological system in
which the production of entropy per unit of information is minim-
ized. Such process consists in substituting biological com-
ponents of the system with other biological components so as to
preserve the same or more information at the same or less ener-
getic cost. Paradoxically, the system seeks to gain information
from the environment only to use such information to block any
further assimilation of information. During succession there is
trend towards increase in biomass, complexity stratification, and
diversity. The more entropy/energy efficient systems are those
that are best fit to survive. Therefore, succession is to ecology
what evolution is to biology.
Marr David: VISION (MIT Press, 1982)
Marr thinks that the vision system employs innate information to
decipher the ambigous signals that it perceives from the world.
Processing of perceptual data is performed by "modules", each
specialized in some function, which are controlled by a central
module. In a similar fashion to Chomsky and Fodor the brain con-
tains semantic representations (in particular a grammar) which
are innate and universal (of biological nature) in the form of
modules that are automatically activated and all concepts can be
decomposed in such semantic representations. The processing of
such semantic representations is purely syntactic.
The physical signal sent to the world is received (in the form of
physical energy) by transductors, which transform it into a sym-
bol (in the form of a neural code) and pass it on to the input
modules, which extract information and send it to the central
module in charge of higher cognitive tasks.
Each module corresponds to neural subsystems in the brain. The
central module exhibits the property of being "isotropic" (able
to build hypotheses based on any other available function) and
"quinian" (the degree of confirmation assigned to an hypothesis
is conditioned by the entire system of beliefs).
The visual system is decomposed in a number of independent
subsystems. Such subsystems provide a representation of the
visual scene at three different levels of abstraction: the "pri-
mal sketch", which is a symbolic representation from the meaning-
ful features of the image (anything causing sudden discontinui-
ties in light intensity, such as boundaries, contours, textures);
a 2 and a half dimensional sketch, which is a representation cen-
tered on the visual system of the observer (e.g., describes the
surrounding surfaces and their properties, mainly distances and
orientation) and computed by a set of modules specialized in
parameters of motion, shape, color, etc; and finally the tri-
dimensional representation, which is centered on the object and
is computed by Ullman's correspondence rules.
Marr thinks that one can be at either at three levels of
analysis: the computational level (which mathematical function
the system must compute, i.e. an account of human competence),
the algorithmic level (which algorithm must be used, i.e. an
account of human performance) and the physical level (which
mechanism must implement the algorithm). Cognitive science should
investigate the mind at the computational level.
Martin James: A COMPUTATIONAL MODEL OF METAPHOR INTERPRETATION
(Academic Press, 1990)
Martin does not believe that the process of comprehending a meta-
phor is a process of reasoning by analogy. A metaphor is simply a
linguistic convention within a linguistic community, an "abbrevi-
ation" for a concept that would otherwise require too many words.
There is no need for transfer of properties from one concept to
another.
A number of Lakoff-style primitive classes of metaphors (meta-
phors that are part of the knowledge of language) are used to
build all the others. A metaphor is therefore built and
comprehended just like any other lexical entity.
Martin-Lof Per: INTUITIONISTIC TYPE THEORY (Bibliopolis, 1984)
The theory of types is an application of intuitionistic logic. It
provides a framework in which to implement the tasks of program
specification, program construction and program verification.
Expressions are built up from variables and constants by applica-
tion and functional abstraction. The meaning of an expression is
provided by a rule of computation. The mechanical procedure of
computing the value of an expression is its "evaluation". The
statement "a is an element of A" can be understood as "a is a
proof of proposition A" or "a is a program for the solution of
A". The specification of a program is a type definition and the
program itself can be derived formally as a proof.
Mason Stephen: CHEMICAL EVOLUTION (Clarendon Press, 1991)
Mason attempts an explanation of the origin of the elements,
molecules and living systems. His theory is close to Julius Rebek
and Stanley Miller, who are trying to create molecules that
behave like living organisms.
Matthews Robert: LEARNABILITY AND LINGUISTIC THEORY (Kluwer
Academics, 1989)
A collection of papers that cover the relations between learning
theory and natural language from Gold's "identification in the
limit" to Osherson's proof that the class of natural language is
finite.
Maturana Humberto: AUTOPOIESIS AND COGNITION (Reidel, 1980)
The book contains two landmark essays: "Biology of cognition" and
"Autopoiesis".
Maturana argues that the relation with the environment molds the
configuration of a cognitive system. Autopoiesis is the process
by which an organism can continously reorganize its own struc-
ture. Adaptation consists in regenerating the organism's struc-
ture so that its relationship to the environment remains con-
stant. An organism is therefore a structure capable to respond
to the environment, and the stimulus is the part of environment
that is absorbed by the structure.
Living systems are units of interaction. They only exist in an
environment. They cannot be understood independently of their
environment. They exhibit exergonic metabolism, which provides
energy for endergonic synthesis of polymers, i.e. for growth and
replication. The circular organization of living organisms con-
stitutes a homeostatic system whose function is to maintain this
very same circular organization. It is such circular organization
that makes a living system a unit of interaction. At the same
time it is this circular organization that helps maintain the
organism's identity through its interactions with the environ-
ment. Due to this circular organization, a living system is a
self-referring system.
Cognition is biological in the sense that the cognitive domain of
an organism is defined by its interactions with the environment.
A living system operates as an inductive system and in a predic-
tive manner: what its organization reflects regularities in the
environment ("what happened once will occur again"). Living sys-
tems are cognitive systems. Living is a process of cognition. The
internal state of a living organism is changed by a cognitive
interaction in a way relevant to the circularity of its organiza-
tion (to its identity). The nervous system enables a broader way
to interact and eventually self-consciousness.
Maturana assumes that intelligent behavior originates in
extremely simple processes: the living cell is nothing special,
but many living cells one next to the other become a complex sys-
tem thanks to autopoiesis.
An autopoietic system is a network of transformation and destruc-
tion processes whose components interact to continously regen-
erate the network An autopoietic system holds costant its organi-
zation (its identity). Autopoiesis generates a structural cou-
pling with the environment: the structure of the nervous system
of an organism generates patterns of activity that are triggered
by perturbations from the environment and that contribute to the
continuing autopoiesis of the organism. Autopoiesis is necessary
and sufficient to characterize a living system.
All living systems are cognitive systems. Cognition is simply the
process of maintaining itself by acting in the environment.
Language is connotative and not denotative. Its function is to
orient the organism within its cognitive domain. Maturana
extends the term "linguistic" to any mutually generated domain of
interactions (any "consensual domain").
Maturana assumes that multi-cellular organisms are born when two
or more autopoietic units engage in an interaction that takes
place more often than any of the interactions of each unit with
the rest of the environment (a "structural coupling"). Inert ele-
ments become macromoleculs, and macromolecules become organic
cells, and so on towards cellular organisms and intelligent
beings.
The structures that are effectively built are those that make
sense in the environment.
Cognition is a purely biological phenomenon. Organisms do not use
any representational structures, but their intelligent behavior
is due only to the continous change in their nervous system as
induced by perception. Intelligence is action. Memory is not an
abstract entity but simply the ability to recreate the behavior
that best couples with a recurring situation within the environ-
ment.
Even human society as a whole operates as a homeostatic system.
Maturana Humberto & Varela Francisco: THE TREE OF KNOWLEDGE
(Shambhala, 1992)
A popular introduction to Maturana's biology of cognition, cen-
tered around the concept that action and cognition cannot be
separated: "all doing is knowing and all knowing is doing". A
living organism is defined by the fact that its organization
makes it continually self-producing (autopoietic), i.e. not only
autonomous but also self-referring ("the being and doing of an
autopoietic system are inseparable"). Life's origins is not a
mystery: at some point of its history the Earth presented condi-
tions that made the formation of autopoietic systems almost inev-
itable. The whole process of life depends not on the components
of a living organism, but on its organization. Autopoiesis is
about organization, not about the nature of the components.
Basic concepts are defined: replication as a process that gen-
erates unities of the same class, copy as a process that gen-
erates an identical unity, reproduction as a process that gen-
erates two unities of the same class, ontogeny as the history of
structural change in a unity that preserves its organization.
Since ontogeny always happens in an environment, the organism has
to use the environment as a medium to realize its autopoiesis.
There occurs a "structural coupling" between a unity and its
environment.
Evolution is a natural drift, a consequence of the conservation
of autopoiesis and adaptation. There is no need for an external
guiding force to direct evolution. All is needed is conservation
of identity and capacity for reproduction.
The nervous system enables the living organism to expand the set
of possible internal states ant to expand the possible ways of
structural coupling.
When two or more living organisms interact recurrently, they gen-
erate a social coupling. Language emerges from such social cou-
pling. Language is a necessary condition for self-consciousness.
Consciousness therefore belongs to the realm of social coupling.
Mayr Ernst: POPULATION, SPECIES AND EVOLUTION (Harvard Univ
Press, 1970)
Mayr surveys the history of evolutionary theories and current
evolutionary research. The modern synthesis can be summarized as:
evolution is due to "the production of variation and the sorting
of variants by natural selection".
Mayr focuses on the biological properties of species and then
deals with population variation and genetics ("phenotypes are
produced by genotypes interacting with the environment", and
genotypes are produced by the recombination of genes of a local
population).
Mayr focuses on variation ("the study of variation is the study
of populations"). All populations contain enough genetic varia-
tion to fuel evolutionary change. Variation in turn poses prob-
lems for adaptation and speciation. Mayr explain the genetics of
speciation by downplaying the role of geographic isolation and
emphasizing and emphasizing the genetic reconstruction of
populations.
The species are the units of evolution. Speciation is the method
by which evolution advances.
The structure of an organism necessarily reflects its evolution-
ary history.
Maynard Smith John: EVOLUTIONARY GENETICS (Oxford University
Press, 1989)
Maynard Smith also faced the paradox of the origin of life. Even
if one assumes that self-replicating entities came to exist spon-
taneously, growing a body was almost impossible: long strings of
RNA need enzyme assistants in order to replicate, but specifying
these assistants requires long strings of RNA... One cannot
exist without the other already existing.
Maynard Smith John: THEORY OF EVOLUTION (Cambridge University
Press, 1993)
Smith is a darwinist who tried to define progress in evolution as
"an increase of information transmitted between generations".
Smith believes that evolution was completely random: if played
again, evolution may lead to completely different beings.
Mayr Ernst: THE GROWTH OF BIOLOGICAL THOUGHT (Harvard Univ
Press, 1982)
A monumental history of evolutionary biology, from Aristotle to
Darwin, from Mendel to the DNA.
Mayr Ernst: TOWARDS A NEW PHILOSOPHY OF BIOLOGY (Harvard Univ
Press, 1988)
In this collection os essays Mayr tackles biological themes from
a philosophical standpoint. Mayr debates extraterrestrial intel-
ligent life, speciation, punctuated equilibria, etc.
Mayr reiterates that the genes are not the units of evolution.
McClelland James & Rumelhart David: PARALLEL DISTRIBUTED PRO-
CESSING vol. 2 (MIT Press, 1986)
The second volume of the seminal collection of articles deals
with psychological processes (thought, learning, reading,
speech) and biological mechanisms (plausible models of neural
behavior) in the light of connectionism.
Paul Smolensky attempts to bridge the symbolic level of cognitive
science and the subsymbolic level of neurosciences.
McGinn Colin: THE PROBLEM OF CONSCIOUSNESS (Oxford Univ Press,
1991)
Consciousness cannot be explained by humans, but only by an
external being, because consciousness does not belong to the
"cognitive closure" of the human organism.
McNeill David: HAND AND MIND (Univ of Chicago Press, 1992)
Following Adam Kenton, McNeill presents a unified theory of
speech and gestures, according to which gestures are an integral
part of language.
Gestures directly transfer mental images to visible forms, con-
veying ideas that language cannot always express. Gestures con-
tribute directly to the semantics and pragmatics of language.
Gestures transform mental images into visual form and therefore
express more than spoken language can express; and, symmetri-
cally, they build in the listener's mind mental images that spo-
ken language alone could not build. Gestures complement words in
that they represent the individual's personal context and words
carry this context to the level of social conventions. Unlike
words, gestures are synthetic, noncombinatorial and never
hierarchical: they present meaning complexities without undergo-
ing the kind of (linear and hierarchical) decomposition that spo-
ken language undergoes. Gestures provide a holistic and imagis-
tic kind of representation, while speech provides a analytic and
linguistic representation. Speech and gesture arise from the
interaction (dialectic) of imagistic and linguistic mental opera-
tions through a process of self-organization.
The book offers a classification of gestures (including meta-
phoric gestures) and a narrative theory of gestures.
McNeill David: PSYCHOLINGUISTICS (Harper & Row, 1987)
The main thesis is that Saussure's linguistic paradigm (language
as a system of static contrasts on the social level, i.e. langue
vs parole, signifier vs signified, synchronic vs diachronic, syn-
tagmatic vs paradigmatic, linguistic value vs intrinsic value)
and Vygotsky's psychological paradigm (language as a dynamic pro-
cess on the individual level) can be reconciled by positioning
them at different points on the speech developmental time axis
(the time it takes to think and build the sentence, not the time
it takes to utter it).
The book contains a clear introduction to Saussure's linguistics.
McNeill's methodology relies on gesture ("gestures are part of
the sentence") as an additional source of evidence.
McNeill also offers his own theory of spontaneous speech genera-
tion: inner speech symbols self-activate in appropriate concep-
tual situations and generate speech. He recognizes two fundamen-
tal types of thinking, and assumes that during linguistic actions
imagistic thinking is unpacked by syntactic thinking.
Linguistic actions create self-aware consciousness: an individual
becomes self-conscious by mentally simulating social experience.
Individual consciousness is social.
Mead George Herbert: MIND, SELF AND SOCIETY (Univ of Chicago
Press, 1934)
Mind and consciousness are products of socialization among bio-
logical organisms. Language provides the medium for their emer-
gence. The mind is therefore socially constructed. Society con-
stitutes an individual as much as the individual constitutes
society.
The mind emerges through a process of internalization of the
social process of communication: reflecting to oneself the reac-
tion of other individuals to one's gestures. The minded organism
is capable of being an object of communication to itself. Mead
focuses on the role of gestures, which signal the existence of a
symbol (and a meaning) that is being communicated (i.e., recalled
in the other individual), and therefore constitute the building
blocks of language. "A symbol is the stimulus whose response is
given in advance". Meaning is defined by the relation between
the gesture and the subsequent behavior of an organism as indi-
cated to another organism by that gesture. The mechanism of
meaning is therefore present in the social act before the cons-
ciousness of it emerges.
Consciousness is not in the brain, but in the world. It refers to
both the organism and the environment, and cannot be located sim-
ply in either. What is in the brain is the process by which the
self gains and loses consciousness (analogous to pulling down and
raising a window shade).
Mead draws a distinction between the "me" (the self of which we
are mostly aware) and the "I" (the self that is unpredictable).
Metzinger Thomas: CONSCIOUS EXPERIENCE (Springer Verlag, 1996)
A collection of papers on the problem of consciousness. Contribu-
tions by Michael Tye, William Lycan, Daniel Dennett.
Michalski Ryszard, Carbonell Jaime & Mitchell Tom: MACHINE
LEARNING I (Morgan Kaufman, 1983)
A collection of seminal papers on machine learning, including
Michalski's "A theory and methodology of inductive learning" (his
"Star" methodology, i.e. learning as a heuristic search in a
space of symbolic descriptions driven by a incremental process of
specialization and generalization) and "Conceptual clustering",
Carbonell's "Learning by analogy", Mitchell's "Learning by exper-
imentation" (his "version spaces" technique, where a version
space is the partially ordered set of all plausible descriptions
of the heuristic and an incremental process of refinement narrows
down the space to one description).
Doug Lenat surveys his projects of learning by discovery (AM,
Eurisko), Langley reports on BACON.
Michalski Ryszard, Carbonell Jaime & Mitchell Tom: MACHINE
LEARNING II (Morgan Kaufman, 1986)
A second set of articles on machine learning research. Includes
reports from Patrick Winston, Thomas Dietterich, Paul Utgoff,
Ross Quinlan, Michael Lebowitz, Yves Kodratoff, Gerald DeJong.
Included are contributions from cognitive architectures (Paul
Rosenbloom, John Anderson), qualitative physics (Kenneth Forbus),
genetic algorithms (John Holland).
Carbonell's analogical reasoning includes "trasformational" rea-
soning (that transfers properties from a situation to another
situation) and "derivational" reasoning (that derives the proper-
ties of a situation from another situation).
Michalski Ryszard & Kodratoff Yves: MACHINE LEARNING III (Morgan
Kaufman, 1990)
A third set of articles that reports on new developments from the
main protagonists of the field.
Michalski Ryszard: MACHINE LEARNING IV (Morgan Kaufman, 1994)
New developments in machine learning, with a section on theory
revision.
Miller George Armitage & Johnson-Laird Philip: LANGUAGE AND PER-
CEPTION (Cambridge Univ Press, 1976)
This monumental book, from a wealth of psychological investigan-
tions of a number of perceptual phenomena, attempts a psychologi-
cal study of the lexical component of language.
"Sense" has two meanings, one perceptual and the other linguis-
tic. The relation between perceptual and linguistic structures is
mediated by a complex conceptual system: percepts and words are
just channels to enter and exit this complex system. Labels are
learned not by pure association, but through an attentional-
judgmental abstraction of perception. We don't learn automatic
links between percepts and words, we learn rules relating percep-
tual judgments to assertible utterances. The relation between
perception and language consists in learning metalinguistic rules
that specify how perceptual judgments can be used to verify or
falsify sentences. The meaning of a sentence is the way of veri-
fying it.
In Johnson-Laird's procedural semantics, a word's meaning is the
set of conceptual elements that can contribute to build a mental
procedure necessary to comprehend any sentence including that
word. Those elements depend on the relations between the entity
referred by that word and any other entity it can be related to.
Rather than atoms of meanings, we are faced with "fields" of
meaning, each including a number of concepts that are related to
each other. The representation of the mental lexicon handles the
intensional relations between words and their being organized
into semantic fields.
Along the way, the authors review hundreds of cognitive theories
about memory, perception and language.
Millikan Ruth: LANGUAGE, THOUGHT AND OTHER BIOLOGICAL CATEGORIES
(MIT Press, 1987)
Millikan aims for a general theory of "proper functions" that can
be applied to body organs, instinctive behaviors and language
devices (all elements used in verbal communication, from words to
intonation). Such proper functions explain the survival of those
entities, in particular of language devices, and therefore eluci-
date what they "do". Proper functions are related with the his-
tory of a thing, with what it was designed to do. Language dev-
ices survive because they establish a symbiotic relationship
between speakers and hearers. A proper function is a function
that stabilizes the realtionship between a speaker and a hearer
with respect to a language device.
Speaker meaning and sentence meaning are related, but neither can
be used as a base for defining the other.
Millikan then develops a general theory of signs and thoughts.
Intentionality is a natural phenomenon: intentions are members of
proper-function categories (i.e., biological categories) that
have been acquired through an evolutionary process for their sur-
vival value. The intentionality of language can be described
without reference to the speaker's intentions. Representations
are a special class of intentional devices, which include sen-
tences and thoughts: when they perform their proper function,
their referents are identified. Beliefs are representations.
Meaning has three parts: the proper function, Fregean sense and
intension.
Millikan Ruth: WHAT IS BEHAVIOR? (MIT Press, 1991)
Millikan, inspired by Dawkins, believes that, when determining
the function of a biological "system", the "system" must include
more than just the organism, something that extends beyond its
skin. Furthermore, the system often needs the cooperation of
other systems: the immune system can only operate if it is
attacked by viruses.
Mines Robert: ADULT COGNITIVE DEVELOPMENT (Praeger, 1986)
A collection of essays on the subject, including Patricia Arlin's
seminal "Problem finding" and Karen Kitchener's "Reflexive judge-
ment model".
Arlin studies the cognitive developmental process that enables
creativity in art and science, or the emergence of postformal
operational thinking that follows Piaget's traditional stages in
the young adult.
Kitchener assumes that an adult keeps developing his or her cog-
nitive faculties and therefore refining the way decisions are
taken in complex situations. Cognitive development continues for
the entire lifetime.
Minsky Marvin: SEMANTIC INFORMATION PROCESSING (MIT Press, 1968)
A collection of articles about seminal, historical artificial
intelligence systems, including Bertram Raphael's SIR for natural
language understanding and Daniel Bobrow's STUDENT. Also includes
John McCarthy's 1958 article on "Programs with common sense" and
Ross Quillian's 1966 paper on "Semantic memory".
McCarthy proposes to build a program that reasons deductively
from a body of knowledge until it concludes that some actions
ought to be performed; then it adds the results of the actions to
its body of knowledge; and repeats its cycle. McCarthy also
sketches for the first time his situation calculus to represent
temporally limited events as "situations".
Quillian defines of a semantic network as a relational direct
acyclical graph in which nodes represent entities and arcs
represent binary relations between entities.
Minsky Marvin: THE SOCIETY OF MIND (Simon & Schuster, 1985)
The book summarizes all of the cognitive ideas of Minsky, from
frames to K-lines.
In a similar vein to Dennett's homunculi, the cognitive architec-
ture of the society of mind assumes that intelligent behavior is
due to the non-intelligent behavior of a very high number of
agents organized in a bureaucratic hierarchy. The set of their
elementary actions and their communications can produce more and
more complex behavior.
Minsky assumes that a data structure called "K-Line" (Knowledge
Line) records the current activity (all the agents currently
active) when a perception or problem solving task takes place and
that the memory of that event or problem is a process of rebuild-
ing what was active (the agents that were active) in the mind at
that time. Agents are not all attached the same way to K-lines.
Strong connections are made at a certain level of detail, the
"level-band", weaker connections are made at higher and lower
levels. Weakly activated features correspond to assumptions by
default, which stay active only as long as there are no con-
flicts. K-lines connect to K-lines and eventually form societies
of their own.
Minsky defines a frame as a packet of information that help
recognize and understand a scene, represent sterotypical situa-
tions and find shortcuts to ordinary problems.
Memory is a network of frames, one relative to each known con-
cept. Each perception selects a frame (i.e., classifies the
current situation in a category) which then must be adapted to
that perception; and that is equivalent to interpret the situa-
tion and decide which action must be performed. Reasoning is
adpating a frame to a situation. Knowledge imposes coherence to
experience.
The frame offers computational advantages (because it focuses
reasoning on the information that is relevant to the situation),
is biologically plausible (it does not separate cognitive
phenomena such as perception, recognition, reasoning, understand-
ing and memory).
A frame is the description of a category by means of a prototypi-
cal member (i.e., its properties) and a list of actions that can
be performed on any member of the category. Any other member of
the category can be described by a similar frame that customizes
some properties of the prototype. A prototype is simply a set of
default properties. Default values express a lack of information,
which can be remedied by new information (unlike with classical
logic, which is monotonic).
A frame provides multiple representations of an object: taxonomic
(conjuctions of classification rules), descriptive (conjunction
of propositions of the default values) and functional (a proposi-
tion on the admissible predicates).
Minsky Marvin: PERCEPTRONS; AN INTRODUCTION TO COMPUTATIONAL
GEOMETRY (MIT Press, 1969)
The book that virtually delivered a near-fatal blow to research
on neural networks by exposing mathematically the limitations of
perceptrons.
Minksy Marvin: COMPUTATION (Prentice-Hall, 1967)
Minsky built a computational connectionist theory on top of
McCulloch's and Pitts' binary neuron.
Mitchell Melanie: ANALOGY-MAKING AS PERCEPTION (MIT Press, 1993)
The book describes the program built by the author and Douglas
Hofstadter, COPYCAT.
Analogy making is viewed as a perceptual process, rather than a
purely reasoning process. The interaction of perceptions and con-
cepts gives rise to analogies.
The computer model entails a large number of parallel processors,
halfway between connectionist and symbolic systems. Concepts and
perceptions are not well defined entities, but dymanic processes
that arise from such a configuration. The system employs an
innovative stochastic search to find solutions.
Mitchell Melanie: INTRODUCTION TO GENETIC ALGORITHMS (MIT Press,
1996)
A brief survey of the field.
Monod Jacques: CHANCE AND NECESSITY (Knopf, 1971)
English edition of "Le hasard et la necessite'".
Based on probabilities, Monod analyzes the interplay of chance
and natural selection in the evolution of life and concludes that
life was born by accident; then Darwin's natural selection made
it evolve. The origin of biological information is also
inherently determined by chance: there is no causal connection
between the syntactic (genetic) information and the semantic
(phenotypic) information that results from it, as the former and
its effects are completely independent of one another.
Monod deals with the paradox that a mono-dimensional structure
like the genome can specify the function of a three-dimensional
structure like the body: the function of a protein is underspeci-
fied in the code, but the environment of the protein determines a
unique interpretation.
Montague Richard: FORMAL PHILOSOPHY (Yale University Press,
1974)
Montague developed an intensional logic that employs a type
hierarchy, higher-order quantification, lambda abstraction for
all types, tenses and modal operators. Its model theory is based
on coordinate semantics.
Reality consists of two truth values, a set of entities, a set of
possible worlds and a set of points in time. A function space is
constructed inductively from these elementary objects.
The sense of an expression is supposed to determine its refer-
ence. The intensional logic makes explicit the mechanism by which
this can happen. The logic determines the possible sorts of
functions from possible indices (sets of worlds, times, speakers,
etc) to their denotations (or extensions). These functions
represent the sense of the expression.
In other words sentences denote extensions in the real world. The
denotation is compositional, meaning that a subpart of the inten-
sion extends or delimits the extension denotated by another.
A name denotes the infinite set of properties of its reference.
Common nouns, adjectives and intransitive verbs denote sets of
individual concepts and their intensions are the properties
necessarily shared by all those individuals.
Montague's semantics is truth conditional (to know the meaning of
a sentence is to know what the world must be for the sentence to
be true, the meaning of a sentence is the set of its truth condi-
tions), model theoretic (a way to carry out the program of
truth-conditional semantics that involves building models of the
world which yield interpretations of the language) and uses pos-
sible worlds (the meaning of a sentence depends not just on the
world as it is but on the world as it might be, i.e. on other
possible worlds).
Montague used his intensional logic to derive a semantic, model-
theoretic interpretation of a fragment of the english language:
through a rigorously mechanical process, a sentence of natural
language is translated into an expression of the intensional
logic and the model-theoretic interpretation of this expression
serves as the interpretation of the sentence.
Montague relalized that categorial grammars provide a unity of
syntactic and semantic analyses.
Rather than proving a semantic interpretation directly on syntac-
tic structures, Montague provides the semantic interpretation of
a sentence by showing how to translate it into formulas of inten-
sional logic and how to interpret semantically all formulas of
that logic. Montague assigns a set of basic expressions to each
category and then defines 17 syntactic rules to combine them to
form complex phrases. An analysis tree shows graphically how a
meaningful expression is constructed from basic expressions. The
tree shows all applications of syntactic rules down to the level
of basic expressions. The translation from natural language to
intensional logic is then performed by employing a set of 17
translation rules that correspond to the syntactic rules. Syntac-
tic structure determines semantic interpretation. The semantics
of the intensional logic is given as a possible-world semantics
relative to moments of time: "points of reference" (pairs of
worlds and moments) determine the extensions of expressions whose
meanings are intensions.
Montague believes there should be no theoretical difference
between natural languages and artificial languages of logicians.
A universal grammar is a mathematical framework capable of sub-
suming a description of any system that might be considered as a
language.
Moore Robert: LOGIC AND KNOWLEDGE REPRESENTATION (CSLI, 1995)
Collects Moore's writings on "knowledge and action", belief
theory and autoepistemic logic.
Morowitz Harold: ENERGY FLOW IN BIOLOGY (Academic Press, 1968)
The thesis of the book is that the flow of energy through a sys-
tem acts to organize the system. The apparent paradox between the
second law of thermodynamics (the universe tends towards increas-
ing disorder) and biological evolution (life tends towards
increasing organization) is solved by realizing that thermodynam-
ics applies to systems that are approaching equilibrium (either
adiabatic, i.e. isolated, or isothermal), whereas natural systems
are usually subject to flows of energy/matter to or from other
systems. Steady-state systems (where the inflow and the outflow
balance each other) are particular cases of nonequilibrium sys-
tems.
Schrodinger's vision that "living organisms feed upon negative
entropy" (they attract negative entropy in order to compensate
for the entropy increase they create by living) can be restated
as: the existence of a living organism depends on increasing the
entropy of the rest of the universe.
Morowitz Harold: FOUNDATIONS OF BIOENERGETICS (Academic Press,
1978)
A classic textbook that introduces thermodynamic concepts
(energy, temperature, entropy and information) and the laws of
statistical mechanics and then applies them to biological struc-
tures such as the solar radiation. Nonequilibrium (irreversible)
thermodynamics is introduced.
Morowitz's theorem states that the flow of energy through a sys-
tem leads to cycling in that system. The flux of energy is the
organizing factor in a dissipative system. When energy flows in
a system from a higher kinetic temperature, the upper energy lev-
els of the system become occupied and take a finite time to decay
into thermal modes. During this period energy is stored at a
higher free energy than at equilibrium state. Systems of complex
structures can store large amounts of energy and achieve a high
amount of internal order.
Therefore, a dissipative system develops an internal order with a
stored free energy that is stable, has a lower internal entropy
and resides some distance from thermostatic equilibrium. Further-
more, a dissipative system selects stable states with the largest
possible stored energy.
The cyclic nature of dissipative systems can be seen in the
periodic attractors. Their cyclic nature allows them to develop
stability and structure within themselves.
Morowitz Harold: ENTROPY AND THE MAGIC FLUTE (Oxford University
Press, 1993)
A collection of short and very entertaining essays on intriguing
topics.
Morris C.W.: FOUNDATIONS OF THE THEORY OF SIGNS (University Of
Chicago Press, 1938)
Morris revised Peirce's theory of signs and introduced the modern
terminology.
Myers Terry: REASONING AND DISCOURSE PROCESSING (Academic Press,
1986)
A collection of papers on discourse structure and analysis.
Includes Johnson-Laird's "Reasoning without logic", a critique of
mental logic, and Wilson's and Sperber's "Inference and implica-
ture in utterance interpretation", on their theory of relevance.
Nagel Thomas: MORTAL QUESTIONS (Cambridge Univ Press, 1979)
Contains the famous "What is it like to be a bat": we can learn
all about the brain mechanism of a bat's sonar system without
having the slightest idea of what it is like to have the sonar
experiences of a bat.
Nagel Thomas: THE VIEW FROM NOWHERE (Oxford Univ Press, 1986)
"We can conceive of things only as they appear to us and never as
they are in themselves." We can only experience how it feels to
be ourselves. We can never experience how it feels to be some-
thing else, for the simple reason that we are not something else.
As Nagel wrote in a famous paper, we can learn all about the
brain mechanisms of a bat's sonar system without having the
slightest idea of what it is like to have the sonar experiences
of a bat.
Nagel stresses the possibility that the human brain may be inade-
quate to fully understand the world and therefore itself.
Therefore we can never be sure that our viewpoint is an objective
viewpoint. We can only be sure that our viewpoint is definitely a
very subjective viewpoint. At the same time we should not limit
ourselves to objective reality, as the subjectivity of conscious-
ness is part of reality. A fully objective view of the world
would omit the fact that, of all things, the world contains "me",
my self, and it is through this self that I observe the world.
Demonstrative words such as "I", "here" and "now" cannot be
reduced to objective reality.
Nagel thinks that identity comes from the (physical) brain, since
it is the only part of a body that a person would not survive
without. Brain transplants would be inherently different from
heart or kidney transplants. If a region of the brain does not
perform its function properly and a region can be found in
another brain that would perform that function properly, can we
transplant it and still have the same person? Nagel does not
answer this question.
Consciousness cannot be "counted": schizophrenic patients have
neither one nor two consciousnesses. Brain emispheres cannot com-
pete, even when they have been separated. They have been pro-
grammed to work in tandem.
Neal Stephen: DESCRIPTIONS (MIT Press, 1990)
Neal's semantic theory is based on Bertrand Russell's theory of
descriptions.
Neisser Ulric: COGNITION AND REALITY (Freeman, 1975)
Cognition is defined as the skill of dealing with knowledge. Such
knowledge comes from the environment. Rather than trying to
understand which information the mind can process, attention
should be paid to which information is available in the environ-
ment. The mind developed to cope with that information.
Neisser, in partial agreement with Gibson, presents an alterna-
tive approach to perception which is based on an ecological
orientation. Organisms pick up information from the environment.
Neisser differs from Gibson in that he argues in favor of direc-
tionality of exploration by the organism: the organism is not
completely passive in the hands of the environment, but somehow
it has a cognitive apparatus that directs its search for informa-
tion.
Schemata (analogous to Selz's and Bartlett's schemata) account
for how the organism can gather the available information from
the environment. Between perception and action there exists a
direct relation. The schema accounts for adaptive behavior while
conserving the preminence of cognitive processes. The organism
selects information from the environment based on its anticipa-
tory schemata. "We can see only what we know how to look for". At
every istant the organism constructs anticipations of information
that enable it to pick it up when it becomes available. Once
picked up, information may in turn result in a change of the ori-
ginal schema, to direct further exploration of the environment.
Perception is therefore a perennial cycle, from schemata to
action (schemata directs action) to information (action picks up
information) to schemata (information modifies schemata). Sche-
mata are part of the nervous system.
Perception of meaning also depends on schematic control of infor-
mation pickup. Perception is not about classifying objects in
categories.
The cyclical theory of perception also explains how the mind
"filters" the huge amount of information that would exceed its
capacity.
An orienting schema of the nearby environment, or "cognitive
map", guides the organism around the environment. A cognitive map
contains schemata of the objects in the environment and spatial
relations between the objects.
Even mental imagery is reduced to perceptual anticipations for
picking up useful information. Images are plans for obtaining
information.
Perception and cognition transform the perceiver: an organism
"is" the cognitive acts it engages in.
Neisser Ulric: MEMORY OBSERVED (Freeman, 1982)
A collection of historical papers on psychological research about
memory. Included are Freud, Luria, Stern, Bateson, etc.
Neisser Ulric: CONCEPTS AND CONCEPTUAL DEVELOPMENT (Cambridge
University Press, 1987)
Neisser identifies five kinds of self-knowledge: the ecological
self (situated in the environment), the "interpersonal self"
(situated in the society of selves), both based on perception,
the private self, the conceptual self and the narrative self.
Neisser Ulric: CONCEPTS RECONSIDERED (Cambridge Univ Press,
19##)
A collection of papers, including Barsalou's 1987 "The instabil-
ity of graded structures", which proved that concepts are not
stable structures (concepts are built on the fly, given the con-
text, and each instance can be quite different from the previous
one).
Neisser Ulric: THE REMEMBERING SELF (Cambridge University
Press, 1994)
A collection of articles on the "narrative self", the fact that
human beings remember what happened to them. Remembering is a
skill that must be learned. Thus, the remembering self must have
a development of its own.
Jerome Bruner believes in a multiplicity of narratives. There is
not a single, static remembered self. What we remember is influ-
enced by social and cultural factors. Self-narratives don't even
depend so much on memory as on thinking. "Self is a perpetually
rewritten story".
Neisser Ulric: THE PERCEIVED SELF (Cambridge Univ Press, 1994)
A collection of articles from distinguished authors on the "eco-
logical self" (situated in the environment) and the "interper-
sonal self" (situated in the society of selves).
Nelson Raymond: LOGIC OF MIND (Kluwer Academics, 1989)
The book presents a comprehensive and ambitious theory of the
mind as a computational system made of rules that are embodied in
the nervous system.
Nerve networks, grammars and cognitive systems are all reduced to
automata (systems of computational rules). Nelson defends the
view that computers and minds are the same type of automaton,
especially against the misapplication of Godel's theorem.
A theory of belief and action is also developed using abstract
Turing machines and automata models. Reference derives from
intentionality, and intentions can be reduced to ways of comput-
ing (based on the idea of partial recursive functions). The
intentional features of the mind can therefore be explained in
mathematical terms. Nelson builds a "logic of acceptance" to
deal with perception: a stimulus pattern is perceived if it is
accepted by the perceiver as a given type, and this process
depends on the perceiver's expections (i.e., perceptual belief is
fullfilled expectation). Desire is then defined in terms of
belief and action. Both belief and desire are therefore reduced
to mathematical quantities, and ultimately to automata computa-
tion.
A theory of truth for a language corresponding to perceptual
belief and a theory of meaning (based on recursive functions) are
worked out.
Newell Allen & Simon Henry: HUMAN PROBLEM SOLVING (Prentice-
Hall, 1972)
A gigantic study on human behavior from the point of view of
information processing and one of the milestones of cognitive
science.
By conducting experiments, the authors concluded that problem
solving involves a mental search through a problem space of pos-
sible solutions in which each step is guided by rules of thumb,
or heuristics.
A problem space consists of a set of knowledge states, a set of
operators on knowledge states, the initial state, the desired
final state. Problem solving takes place by search in the problem
space until the desired knowledge state is achieved. Knowledge
about the environment is fundamental in order to guarantee a
highly selective search through the problem space.
As an example, the Logic Theorist is a heuristics-based problem
solver whose task is to find proofs for theorems in the proposi-
tional calculus. The General Problem Solver is an even more
ambitious program.
The cognitive model is one in which human intelligence is due to
a set of production rules controlling behavior and to internal
information processing. Both the mind and the computer are
physical-symbol systems.
Newell Allen & Rosenbloom Paul: THE SOAR PAPERS (MIT Press,
1993)
A collection of papers on the unified cognitive architecture
developed over a decade by Rosenbloom, John Laird and Allen
Newell that attempts to explain how a cognitive system can
improve its skills through experience.
The universal weak method is an organizational framework whereby
knowledge determines the weak methods employed to solve the prob-
lem, i.e. knowledge controls the behavior of the rational agent.
Universal subgoaling is a scheme whereby goals can be created
automatically to deal with the difficulties that the rational
agent encounters during problem solving.
The engine of the architecture is driven by production rules that
fire in parallel and represent task-dependent knowledge. The
architecture maintains a context which is made of four slots:
goal, problem space, state and operator. A fixed set of produc-
tion rules determines which objects have to become current, i.e.
fill those slots. In other words, they determine the strategic
choices to be made after each round of parallel processing.
Newell Allen: UNIFIED THEORIES OF COGNITION (Harvard Univ Press,
1990)
Newell divides cognition into several levels. The program level
represents and manipulates the world in the form of symbols. The
knowledge level is built on top of the symbolic level and is the
level of rational agents: an agent has a body of knowledge, some
goals to achieve and some actions that it can perform. An agent's
behavior is determined by the "principle of rationality": the
agent performs those actions that, on the basis of the knowledge
it has, bring it closer to the goals.
General intelligent behavior requires symbol-level systems and
knowledge-level systems.
Newell then broadens his division of cognitive levels by includ-
ing physical and biological states. The whole band can be divided
into four bands: neural, cognitive, rational and social. The cog-
nitive band can be divided based on the response times: at the
memory level the response time (the time required to retrieve the
referent of a symbol) is about ten milliseconds; at the decision
level the response time is 100 milliseconds (the time required to
manipulate knowledge), at the compositional level is one second
(time required to build actions), a tthe execution level ten
seconds (time required to perform the action).
In the rational band the system appears as a goal-driven organ-
ism, capable of processing knowledge and of exhibiting adaptive
behavior.
Newell surveys a number of cognitive theories and cognitive
architectures, particularly SOAR, which is offered as a candidate
for a unified theory of cognition.
Nicolis Gregoire & Prigogine Ilya: SELF-ORGANIZATION IN NONE-
QUILIBRIUM SYSTEMS (Wiley, 1977)
A milestone and monumental work that redefined the way scientists
approach natural phenomena and brought self-organizing processes
to the forefront of the study of complex systems such as biologi-
cal and social ones.
The book introduces nonequilibrium thermodynamics, which leads to
bifurcation theory and to the stochastic approach to
fluctuations.
Under special circumstances the distance from equilibrium and the
nonlinearity of a system become sources of order, driving the
system to ordered configurations (or "dissipative structures").
In dissipative structures nonequilibrium becomes a source of
order.
The multiplicity of solutions in nonlinear systems can be inter-
preted as a process of gradual "emancipation" from the environ-
ment.
A stunning number and variety of fields of application, from
chemistry to sociology. In this framework the most difficult
problems of biology, from morphogenesis to evolution, find a
natural model. A thermodynamics of evolution and even equations
for ecosystems are proposed.
Nicolis Gregoire & Prigogine Ilya: EXPLORING COMPLEXITY
(W.H.Freeman, 1989)
An introduction to the theory of dynamical systems. After provid-
ing examples of self-organization in chemical, cosmological and
biological systems, systems are partitioned into conservative
systems (which are governed by conservation laws for energy,
translational momentum and angular momentum, and give rise to
reversible processes) and dissipative systems (which give rise to
irreversible processes). Equilibrium states and nonequilibrium
constraints are defined operationally, with the emphasis on
fluxex between a system and the environment. A system subject to
the action of a nonequilibrium constraint becomes susceptible to
change as localized tendencies to deviate from equilibrium are
amplified, thus becoming sources of innovation and diversifica-
tion. The potentialities of nonlinearity are dormant at equili-
brium but are revelead by nonequilibrium: multiple solutions
appear and therefore diversification of behavior becomes possi-
ble. Dissipative structures emerge under nonequilibrium condi-
tions. Therefore, nonlinear systems driven away from equilibrium
can generate instabilities that lead to bifurcations and symmetry
breaking beyond bifurcation. The methodology of phase spaces is
introduced to study nonlinear nonequilibrium systems, leading to
formal definitions of limit cycles, attractors, fractals, etc.
Catastrophe and chaos theories are viewed as special cases. A
model of bifurcation and evolution is worked out. The
relationship between stochastic and deterministic behavior
(between chance and necessity) is analyzed, as well as the origin
of irreversibility.
Nicolis Gregoire: INTRODUCTION TO NONLINEAR SCIENCE (Cambridge
University Press, 1995)
The ultimate textbook on nonlinear methods for describing complex
systems. From an interdisciplinary introduction, the book goes
on to introduce in a rigorous manner the vocabulary and tools of
invariant manifolds, attractors, fractals, stability, bifurcation
analysis, normal forms, chaos, Lyapunov exponents, entropies.
Nilsson Nils: THE MATHEMATICAL FOUNDATIONS OF LEARNING MACHINES
(Morgan Kaufmann, 1990)
A revised edition of his seminal 1965 "Learning Machines".
Nilsson Nils: PRINCIPLES OF ARTIFICIAL INTELLIGENCE (Tioga,
1980)
One of the most popular textbooks of artificial intelligence. The
focus is on production systems (heuristic search algorithms,
resolution and unification, planning systems) with a brief men-
tion of semantic networks.
Nunberg Geoffrey: THE PRAGMATICS OF REFERENCE (Indiana Univ
Linguistic Club, 1978)
There's a fundamental ambiguities in all terms: there is always
potentially an infinite number of referents of a term, depending
on the context. Nunberg argues that a term cannot have a stan-
dard referent, but its referents can be derived one from the
other through a number of elementary functions (such as "owner
of" or "location of") which can be recursively applied in any
combination.
Four principles determine which functions a listener is going to
employ to derive the most appropriate referent.
A term is used in a "normal" way when it is consistent with the
conventions of the linguistic community.
A metaphor is a discourse in which the speaker a) employs an
expression E to refer to F in context C even if there exists
another expression to refer to F which the speaker knows it is
easier to understand; b) knows that employing E is not rational
but expects the listener to realize this and that he is aware of
it; c) acts according to a cooperative principle and expects the
listener to be aware of it. Metaphors are not an exclusive of
poets. Quite the opposit: people who are not very fluent in the
language tend to use metaphors more often.