What is Cognitive Science? Lynn Nadel and Massimo Piattelli-Palmarini1 Department of Psychology, University of Arizona, Tucson, AZ

Rich scientific disciplines defy simple definition, ability to visualize fine details of the central ner- and cognitive science is no exception. For our pres- vous system (CNS) settled the debate between the ent purposes, cognitive science can be defined Italian Camillo Golgi (1843±1926) and the Spaniard broadly as the scientific study of minds and brains, Santiago RamoÂn y Cajal (1852±1934) over the nature be they real, artificial, human or animal. In practice, of connection and communication within the brain cognitive science has been more limited, largely (See Golgi Staining; Neuron Doctrine). Golgi had restricting itself to domains in which there is rea- argued that the nervous system was a meshwork sonable hope of attaining real understanding. The of connected elements. Cajal, on the other hand, richness and diversity of the contributions to this argued that there were discrete elements within encyclopedia show that there are now many such the CNS, subsequently called neurons, that were domains. not actually in contact, but instead communicated By `understanding' we mean going beyond across a gap ± the synapse. Stained sections sup- common-sense intuitions, often to the point of rad- ported Cajal, and the era of neurons and synapses ically subverting them. It is no longer surprising began.2,3 The subsequent study of neuronal trans- when a cognitive system is shown to work in highly mission in the squid giant axon and the develop- unexpected ways. One of the insights of modern ment of single neuron recording methods in cognitive science is that the mind often works in vertebrates built upon this early work. Parallel de- counterintuitive ways. velopments in the study of logic and computation Mature sciences owe much of their initial pro- stimulated early attempts to apply mathematical gress to the pursuit of phenomena and hypotheses principles to nervous system function (McCulloch within a few `windows of opportunity', often and Pitts, 1943), and these in turn eventually led to opened by chance. Many of these seem, at their modern approaches in computational neuroscience inception, to be quite far from the daily concerns (See Computational Neuroscience: From Biology to of ordinary people, but they come to have great Cognition). impact. Cognitive science has thrived on such Attempts to ameliorate diseases of the nervous opportunities. As an example, the analysis of lan- system have played a significant role in the history guage impairments following stroke or war injuries of cognitive science. The treatment of recurrent by a group of outstanding neurologists in the nine- epileptic seizures resistant to all drugs led to ex- teenth century (Broca, 1878; Wernicke, 1874) led to perimental surgical treatments that have produced important insights about both the organization fundamental knowledge about brain function (See of language and its instantiation in the brain Epilepsy). The use of radical resection of the tem- (See Aphasia; Broca, Paul; Language Disorders; poral lobe had the unfortunate consequence of Wernicke±Geshwind Model). The use of rapid-suc- causing a massive amnesic syndrome, but the cession photography by the French physiologist study of the patient known as H.M. (Scoville and Marey (1830±1904; Braun, 1992) and by the expatri- Milner, 1957) and other such patients has contrib- ate Englishman Eadweard Muybridge in the U.S. uted greatly to our understanding of memory. (also, by a curious coincidence, born in 1830 and Similarly, the use of surgical section of the corpus deceased in 1904) (Haas, 1976) made possible the callosum led to the discovery of the `split brain', analysis of the natural motion of people and horses which has fascinated philosophers and the public and opened the way to our present understanding at large ever since (Gazzaniga, 1970; Sperry, 1968; of the uniqueness of such motion in perception and See Sperry, Roger; Split-brain Research). planning (See Motion Perception, Neural Basis of; In this general introduction we will briefly ex- Motion Perception, Psychology of). plore these and some other `windows of opportun- The development of highly selective neuron ity' that helped create cognitive science as we know staining methods, which led to the emergence of it today. We cannot be exhaustive but we hope in the `neuron doctrine', is yet another example. The this short introduction to provide a broad roadmap xiv What is Cognitive Science? to the field, where it has come from, and where it by William James of The Principles of Psychology might be headed. We are aware that the names of (James, 1890; also see James, William); (2) the afore- protagonists are often introduced without proper mentioned emergence of the neuron doctrine in the biographical presentation, and that many concepts work of Cajal; and (3) the development of Freud's and terms are briefly characterized or sometimes psychodynamic approach (See Freud, Sigmund). just mentioned. It is our goal here to offer a general All three had profound and lasting effects, al- synopsis, partly historical, partly analytic, but though many might argue that the influence of hopefully sufficient to situate people and ideas in Freud's thinking has waned; time will tell. A fourth a vast web, trusting that the reader will be stimu- strand of intellectual development that has had a lated to locate related articles in this work. Our task profound influence on cognitive science can also be is made somewhat easier by the existence of some traced back to the nineteenth century ± the emer- excellent histories of cognitive science (e.g. Alber- gence of computational devices. Fascinating histor- tazzi, 2001; Baars, 1986; Bechtel et al., 1999; Dupuy, ies have been written about the role of key 2000; Gardner, 1985), as well as numerous reflec- individuals such as Augusta Ada Byron (Stein, tions on history by key participants (to which we 1985) and Charles Babbage in this development, shall refer below; also see History of Cognitive Sci- but real progress in this domain was not seen ence and Computational Modeling). Inevitably, our until the twentieth century. Finally, from philoso- road map will reflect the journeys we ourselves phy and linguistics came the development of have taken; we apologize in advance to those of powerful systems of logic (Frege, 1879; Russell, our colleagues whose contributions to cognitive 1900, 1919; Tarski, 1935, 1996; Whitehead and Rus- science have been overlooked in what follows. sell, 1910; also see Frege, Gottlob). These various advances began to provide the SOME PREHISTORY basis for formal treatments of many aspects of cog- nitive function, and it is these treatments that, in the Interest in mind and brain is as old as recorded fullness of time, combined with germane develop- history, and any complete rendering of the prehis- ments in experimental psychology, neuroscience, tory of cognitive science would treat early philoso- linguistics and anthropology. The convergence of phers at some length. That, however, is not our these strands produced modern cognitive science. purpose. Rather, we will take it for granted that interest in fundamental questions about cognition THE BEHAVIORIST INTERREGNUM and its physical bases has long existed, well before the term `cognitive science' was coined, and pick Notwithstanding the explosion of possibilities up our story at the point where genuinely scientific offered by the developments noted above, the be- analyses became possible in the nineteenth century. ginning of the twentieth century saw a turning An interesting perspective on this gestational era is away from many of the issues central to cognitive offered in Albertazzi (2001). science, especially in North America. The behavior- We can identify several strands of nineteenth cen- ist revolution, exemplified by John B. Watson tury thought as clearly antecedent to modern cogni- (1924), can be viewed in retrospect as a reaction tive science: the work of the neuropsychologists against the overly ambitious reach of early cogni- who studied the impact of brain damage on lan- tive scientists.5 Behaviorists rightly pointed out guage and other cognitive abilities; the development that not enough was known about what goes on by Darwin and others of the theory of evolution inside the organism to ground any sort of mean- (later to be extended into theories of the evolution ingful theory. Instead, behaviorism argued that of the brain and mind; also see Evolutionary Psych- psychology should restrict itself to what could be ology: Applications and Criticisms; Evolutionary observed, which ruled out consciousness and other Psychology: Theoretical Foundations); the creation products of introspection. Combined with an infec- of modern experimental psychology and psycho- tious enthusiasm for spreading `rigorous' scientific physics (Ebbinghaus, Helmholtz and Wundt de- methodology to all fields of inquiry, behaviorism serve special mention here; also see Ebbinghaus, effectively banished all appeals to internal states Hermann; Helmholtz, Hermann von; Wundt, and representations (concepts, ideas, meanings, Wilhelm); and the initial efforts of neurologists and percepts, computations, etc.): better to focus on psychiatrists to relate complex human conditions to what could be observed and measured if one underlying neuroanatomy and neuropathology.4 wanted to create a science.6 Several landmarks at the very end of the nine- A strict behaviorist view, and its emphasis on teenth century stand out: (1) the publication in 1890 general learning mechanisms, persisted for about What is Cognitive Science? xv

50 years in North America, during which time the Wertheimer,KohlerandKoffka,steadfastlyretained center was occupied by narrowly conceived re- a focus on the role of organization in perception search programs, some of which bore considerable and problem solving, with Kohler's work on `in- fruit. Clark Hull's (1943) early efforts at producing sight' being a particularly important contribution. mathematical models of behavior were largely Bartlett's book Remembering (1932) explored the role unsuccessful but they provided the foundation of schemata in the formation of memory, and upon which a more modern and influential math- remains influential to the present day (See Bartlett, ematical psychology was subsequently built (See Frederic Charles). Piaget's (1930, 1954) comprehen- Hull, Clark L.).7 sive studies on development, which emphasized the Three major exceptions to the narrow behaviorist internal models formed by children in comprehend- perspective of this era were Karl Lashley, Edward ing the world, had an enormous impact in Europe, Chace Tolman and Egon Brunswik (See Lashley, and subsequently in North America when they were Karl S.; Tolman, Edward C.). Lashley's (1929) work made widely known after World War II (See Piaget, on neural mechanisms underlying intelligence and Jean).9 Finally, there were the convergent contribu- his thoughts about the localization of function in tions by Kurt GoÈdel, Alan Mathison Turing, Alonzo the nervous system provided a foundation for the Church and Stephen C. Kleene on fundamental subsequent efforts of many influential neuropsy- issues in computation, recursion and the theory chologists of the 1940s and 1950s, including Donald of automata (Church, 1936; Kleene, 1936; Turing, Olding Hebb (Hebb, 1949; also see Hebb, Donald 1936a,b; also see Computability and Computational Olding). Tolman and Brunswik were strongly in- Complexity; Turing, Alan).10 fluenced by the Gestalt movement in Germany, which had focused on the role of higher-order THE MODERN ERA BEGINS factors such as form relations and organization in accounting for perception, insight and even learn- In 1936, Turing published his crucial paper on ing (See Perception, Gestalt Principles of). They set `Computability', in which he spelled out the design the stage for a return of cognitive approaches in for a machine that could carry out any set of well- North America. Tolman's early book, Purposive defined formal operations. Two years later, Shan- Behavior in Animals and Men (1932), provided a non (1938) demonstrated that on±off electrical roadmap for the pursuit of aspects of behavior circuits could carry out basic mathematical pro- that went beyond the observable. Brunswik, an cedures, an idea that ultimately led to the develop- expatriate Viennese, joined the psychology faculty ment of information theory (Shannon and Weaver, of Berkeley in 1937 thanks to Tolman, but remained 1949/1998; also see Information Theory; Shannon, a maverick his entire life (committing suicide in Claude). A critical early contribution was pub- 1955; see Bower, 2002). He introduced the idea of lished by Kenneth Craik (1943), entitled The Nature `ecological validity' (usually associated with the of Explanation. Craik sought ways to link mental much better known work of James Gibson and and mechanical operations, and settled on the Ulric Neisser), defying the relevance of the narrow notion of internal models that would become cen- laboratory settings so dear to the behaviorists. tral to cognitive science in the future. He claimed Brunswik studied the role of sensory cues and sub- (p. 85) that: jective estimates in shaping perception and judg- thought is a term for the conscious working of a highly ment. He advocated the view that knowledge is a complex machine, built of parts having dimensions probability-based process, developed a `probabil- where the classical laws of mechanics are still very istic functionalism', and was among the first to nearly true, and having dimensions where space is, to reveal subjective probability biases. His better all intents and purposes, Euclidean. This mechanism, I known `lens model' pictures systematic distortions have argued, has the power to represent, or parallel, at the interface between the external scene and the certain phenomena in the external world as a calculat- observer, whereby the structure of the environment ing machine can parallel the development of strains in a bridge. is filtered by the structure of subjective perception and knowledge. This results in `perceptual com- According to Craik, thought involved three crit- promises' fit to serve the relevant purposes of the ical steps: first, external processes were translated observer.8 into words, numbers or symbols; second, these Behaviorism was not as dominant in Europe as `representations' were manipulated by processes it was in North America, and elements of what was such as reasoning to yield transformed symbols; to become cognitive science proceeded in a variety and third, these transformed symbols were retrans- of domains. The Gestalt psychologists, including lated into external processes to yield a product, xvi What is Cognitive Science? such as behavior. The critical assumption here is Theory), and his contributions to the development the idea that minds create internal models and then of computers (von Neumann, 1951) were critical at use these models to predict the future. Such a the outset, and he was a major participant in the thought process allows an organism the luxury of emergence of cybernetics until his early death in trying out possible futures before settling on the 1957 (See von Neumann, John). His posthumously one that would be most adaptive. For Craik, published essay on The Computer and the Brain dis- thoughts could not be separated from feelings, a cusses the fundamental properties of computation perspective that early cognitive science ignored to in machine and brain, laying out `an approach its detriment. He died in a bicycling accident a few toward the understanding of the nervous system years after the publication of this work, and further from the mathematician's point of view' (von development depended upon others. Neumann, 1957, p. 1). One of the major historical forces propelling this A powerful additional incentive, though initially development was World War II, which presented a mostly indirect, came from plans to put the new set of military problems that required rapid com- programmable computers and proto-robots (Ashby, putational solution. The breaking of military secret 1960; Walter, 1953; See Walter, Grey) to good use in codes, the calculation of artillery fire trajectories, navigation and learnable self-steering, text transla- and several problems faced by the real-time reac- tion between languages, selective and addressable tions of airplane pilots soon occupied the best archiving, automatic abstracting of documents, minds of the time, and enormous progress in visual detection and discrimination, problem- understanding complex systems resulted. `Cyber- solving and automated induction. The impact of a netics', arguably the most crucial of the disciplines practically-oriented engineering perspective and that paved the way to cognitive science (Dupuy, of generous financial resources on what had pre- 2000), was defined by one of its inventors (the MIT viously been abstract and elusive domains of mathematician Norbert Wiener) as the science of academic research generated some rather naive ap- communication between complex systems (natural proaches and exaggerated expectations, but there or artificial) and of the control of such systems by were also fresh starts, original redefinitions of many intelligent agents. It derives its name from the problems and new models that were unencum- Greek `kybernetis' (the skipper of a boat and, by bered by ancient paralyzing paradoxes and a extension, the pilot of an airplane), betraying its stifling received wisdom. origin in concrete problems posed by the war effort Arguably the best example is to be found in (Wiener, 1948). linguistics, which had been mostly a literary, com- Many of the participants in this fascinating era parative and philological discipline. It suddenly have provided at least some historical record, as received a new impulse to rethink its very founda- noted already, and the interested reader is directed tions and to explore computer-assisted applica- to these sources for a fascinating tour (e.g. McCul- tions. These new approaches to the structure of loch, 1988; Heims, 1991). Almost all agree that a few language (Bar-Hillel, 1954; Chomsky, 1955, 1957, critical concerns fueled the enterprise: first, there 1975; Harris, 1951, 1952a,b, 1957, 1986/1951) were was a strong desire to bring the insights of math- in some measure a reaction to shotgun engineering ematical modeling (the solution of complex differ- attempts to build automatic translators. It is em- ential equations) and of mathematical logic into blematic, for instance, that starting in the mid- contact with both biology and engineering. Two 1950s and for decades to come, the team of linguists of the key players were trained by giants in logic: that gave rise to generative grammar was assem- Norbert Wiener under the guidance of Bertrand bled and then sustained in a school of electrical Russell, and Walter Pitts under Rudolph Carnap. engineering.11 Second, there was a strong commitment to the All in all, this early seminal period proved cru- notion, outlined by Craik, that thought could be cial in engendering the conviction that long-stand- viewed as a computational process utilizing in- ing hard problems in the study of the mind and ternal models, and hence cognition would ultim- brain were open to radically new insights, and that ately be accounted for in terms of finite and an intense collaboration between different scientific specifiable procedures that could be performed by fields (from logic to neurology, from text analysis to the `computers' that were then being developed. In the mathematical theory of recursive functions) this context we must mention John von Neumann, was not only possible, but mandatory. The end of who played a central role in the early days of cog- World War II heralded an era of `visiting' scientists, nitive science. His work on game theory (von mobile young PhDs and ardent interdisciplinary Neumann and Morgenstern, 1944; also see Game exchanges. Not unlike the immense impact that What is Cognitive Science? xvii high-level scientific meetings had previously had Ten Macy Foundation conferences on cybernet- on physics in the 1920s and 1930s, the role that ics were held between 1946 and 1953. The major some meetings in the 1940s and 1950s had in moving force behind these meetings was Warren shaping what later became cognitive science de- McCulloch, as portrayed in considerable detail by serves special attention. Heims (1991) and Dupuy (2000). No written record exists of the first five meetings, but proceedings of THE 1940s: SEMINAL MEETINGS FOR the final five were published. Over the course of COGNITIVE SCIENCE these meetings, ideas central to cybernetics ± such as control, feedback and communication ± were In reconstructing the history of the early years, one imparted to a wide range of scientists. What dis- cannot escape the crucial role played by a number tinguished these meetings was their interdiscipli- of critical meetings at which key participants from narity. The anthropologists Gregory Bateson and several fields were profitably brought together. The Margaret Mead were as central to the proceedings first such meeting, held in 1942 in New York City, as the neurophysiologist McCulloch, the mathem- focused on the topic of central inhibition in the aticians Wiener and von Neumann, the information nervous system. It was sponsored by the Josiah theorist Shannon and the psychologists Kluver Macy Jr. Foundation, which would play an abso- and Lewin (See Lewin, Kurt). Another distinguish- lutely essential role in the birth of cognitive science ing feature of these early meetings on cybernetics over the next decade. At this meeting, Warren was the lack of concern with computation as a McCulloch and Arturo Rosenblueth presented ma- `symbolic' activity. This perspective, which came terial related to the papers they were about to pub- to dominate cognitive science in the 1950s, was lish (McCulloch and Pitts, 1943; Rosenblueth et al., largely absent from these early formative dis- 1943; See McCulloch, Warren). These papers sug- cussions. gested, in rather different ways, that aspects of Some of the same individuals participated in the mental activity could be modeled in a formal way famous Hixon symposium, held at Caltech in 1948, using idealized nervous system elements (an idea and published a few years later (Jeffress, 1951).13 It that was to be revamped many years later by the was at the Hixon meeting that von Neumann gave connectionists, with radically different mathemat- a paper on `The general and logical theory of au- ical models, as we shall see). McCulloch and Pitts tomata', and Lashley wowed the audience with his showed that networks of on±off neurons could famous paper on `Serial order' (Lashley, 1951), in compute logical functions, while Rosenblueth et al. which he took a strong stand against the reigning (1943), concerned with purpose, showed that goal- behaviorist stance, pointing out that the require- directed behavior could emerge in systems with ments of speech and language rendered stimulus± feedback. response theory implausible. In the winter of 1943±44, another meeting was Another critical meeting was held September held at Princeton, attended by McCulloch and Pitts, 10±12, 1956 at MIT ± the second Symposium on by a prominent neuroanatomist, Lorente de No, Information Theory. George Miller has referred to and by two leading figures in the emerging com- the session held on September 11 as the actual birth puter paradigm, John von Neumann and Herman day of cognitive science (but see Wildgen, 2001 for Goldstine. This line-up suggests the coming to- a different perspective). Speaking that day were gether of formal logic, neural networks, real neuro- Newell and Simon, who presented material on anatomy, and computation. Lorente de No (1938) their logic theory machine; Rochester, who had had previously demonstrated, in elegant anatom- been using a large digital computer in a failed effort ical work, that conditions existed within the cere- to test Hebb's theory of cell assemblies (Rochester bral cortex for reverberatory circuits that could et al., 1956); Chomsky, who showed how linguistics instantiate re-entrant loops. These loops were seen could produce results with considerable math- as critical to maintaining memory within the brain, ematical rigor; Miller, who talked about the limits a requirement that had been hinted at earlier by of short-term memory; and Swets, who presented Kubie (1930) and was adopted in Hebb's (1949) on signal detection theory and perceptual recogni- neuropsychological theory of cell assemblies and tion. George Miller later claimed that he went away phase sequences (See Hebbian Cell Assemblies). from this meeting with the feeling that a new sci- Hebb's book The Organization of Behavior stands ence was emerging. out as a crucial contribution, merging psychology Some months before this germinal meeting, a and physiology in pursuit of explanations of cogni- research seminar on artificial intelligence was held tive function.12 at Dartmouth, attended by most of the individuals xviii What is Cognitive Science? active in the area at the time (See Artifical Intelli- reasons: (1) the early biologically driven approach gence, Philosophy of).14 Newell and Simon (1972) to neural nets, as exemplified by the perceptron point out in the Historical Addendum to Human model (Rosenblatt, 1962; See Perceptron) appar- Problem Solving that Minsky's (1961) essay, `Steps ently failed, the limitations of these early models toward Artificial Intelligence' ± first circulated in being subsequently made clear by Minsky and draft form at this summer meeting ± captures the Papert (1969/1990) ± Hanson (1999) provides an consensus views that existed at that time (available interesting perspective on the competition between on Minsky's website at: http://web.media.mit. Rosenblatt and Minsky, and how the latter's views, edu/~minsky/). tilted towards artificial intelligence and away from In meetings such as these one can see the mix of biology, carried the day ± and initial attempts by disciplines that would come to define much of Rochester and his colleagues (Rochester et al., 1956) modern cognitive science. One can also see the to simulate the neurobiology of Hebb and Peter groping for methodologies that would permit sci- Milner were largely unsuccessful, as noted already; entists to ask meaningful questions about mental (2) writers such as Chomsky argued persuasively activity. The single-minded emphasis on behavior that it was the formal properties of the mind±brain at the expense of cognition was clearly at an end. that mattered, not the underlying biology that Within a few years the first of many centers allowed it to compute;16 (3) one of the main repre- of cognitive studies was started, at Harvard, sentatives from neuroscience, Karl Lashley, was by Jerome Bruner (focusing on spontaneous himself rather skeptical about the enterprise. At a reasoning; Bruner et al., 1956) and George Miller symposium on `The Brain and the Mind' at the (focusing on language and memory; Chomsky and American Neurological Association meeting in Miller, 1963; Miller and Chomsky, 1963). Roger 1951, Lashley served as a discussant on several Brown soon transferred there from MIT, enriching papers and had this to say about the enterprise of the research domain with his pioneering studies of linking brains and computers: first-language acquisition (Brown, 1958, 1973). The impact of this center was enormous: its interdiscip- I suggest that we are more likely to find out how the linary mix included faculty, research fellows, visit- brain works by studying the brain itself and the phe- nomena of behavior than by indulging in far-fetched ors and local-area researchers who over the years physical analogies. The similarities in such com- made substantial contributions to cognitive sci- 15 parisons are the product of an oversimplification of ence. Now, over 40 years later, many such pro- the problems of behavior (quoted in Beach et al., 1960, grams exist. p. xix). In the 1950s, as clearly seen in the 1956 MIT meeting, the close connection between biology What is more, Lashley's probabilistic views of and cognitive science fell apart. Simon was perhaps nervous system function were very much at odds the leading, but by no means the only, exponent with the connectionist requirements of the cyber- of the view that to understand cognition one need neticists. Lashley participated in many of the early not pay much if any attention to the underlying critical meetings, and his views had a major impact. biology. Newell and Simon (1972), for example, The rather more positive approach of Hebb, who characterized Hebb's position as `confused' insofar disagreed with Lashley about the role of specific as Hebb thought he was proposing a physiological neural pathways, could not overcome Lashley's account of behavior. In their view it was essential to influence at that time. interpose `a specific layer of explanation lying be- Because of this schism, cognitive science and tween behavior, on the one side, and neurology on neuroscience developed separately after the 1950s. the other' (p. 876). This perspective foreshadowed a For the better part of 25 years, much of neurosci- similar position, presently mainstream, adopted by ence was reductionist in scope and purpose, rarely Marr (1982) a decade later, in his classic treatise on speaking to questions of interest to cognitive scien- vision (See Marr, David). The net result of this shift tists. On the other side, much if not all of cognitive in emphasis was that the integrative perspective of science proceeded within a symbolic framework the cyberneticists was sacrificed, and an era of that required little or no contact with the brain. symbolic modeling, of direct study of mental com- Neurons were relegated to the role of `mere imple- putations and representations, and of artificial in- mentation'. telligence research without much reference to real The then prevailing philosophy of mind, called brains, blossomed. `functionalism' (Dennett, 1987; Fodor, 1975, 1981; It is of historical interest to consider why this Putnam, 1960, 1973) offered principled arguments happened. We would suggest a few possible as to why any physicobiological implementation of What is Cognitive Science? xix cognitive functions was secondary to a thorough selectivity of brain function was shown to reflect abstract characterization of the logical structure of not just exogenous factors, but endogenous ones the mental representations and transformations in- as well. The implications of this were enormous, volved in those functions (See Functionalism).17 as Hebb noted.18 Another critical research program centered EMPIRICAL RESULTS around Penfield and the group of scientists at the Montreal Neurological Institute (MNI), largely The 1950s and beyond witnessed an impressive focused on patients about to undergo surgery to growth in empirical results in all the domains of control for epilepsy (See Penfield, Wilder). Penfield cognitive science. Chomsky revolutionized the and Rasmussen (1949) pioneered the method of study of language; Broadbent (1958) and others stimulating the exposed brain in areas adjacent to focused on attention (See Attention); Bruner and the presumed site of the focus as a means of deter- his colleagues (1956) looked at thinking; Newell, mining which tissue should be excised and which Shaw and Simon (Newell et al., 1958) produced spared. This method yielded several remarkable the General Problem Solver; Hochberg (1956) stud- and widely reported results. First, this method gen- ied the role of memory and other internal factors on erated the famous pictures of sensory and motor perception; Sperling's (1960) work on brief visual maps in the cortex, within which various body presentations and partial report methods led to parts were represented in often unusual propor- the notion of an iconic memory store; and there tions. Second, punctate stimulation in the temporal were various thrusts in artificial intelligence (e.g. lobe could apparently yield the retrieval of a highly Samuel's checkers program, 1959; also see Samuel's specific and detailed memory. This finding strongly Checkers Player), including interesting work on countered Lashley's nonlocalizationist perspective, mathematical neural networks (e.g. Selfridge, and dramatically affected views about the organ- 1958; Rosenblatt, 1958). But there were problems. ization of information in the brain. It seemed to In many cases the successes were garnered in se- promise that an approach depending upon specific verely restricted systems, with no certainty that neural connections might indeed have merit, they would scale-up or generalize. Some domains much as Hebb (Penfield's colleague at McGill, and were simply not part of the mix ± the study of Lashley's former student) had proposed. At the emotion, or consciousness, was ruled off limits same time, the study of H.M., another patient at (See Emotion; Consciousness). During this period, the MNI ± who had lost the capacity to memorize only those phenomena of which humans were recent events, though he maintained some capacity somehow at least partially conscious qualified as to remember events that preceded the surgical bi- `cognitive' ± implicit capacities did not make the lateral section of part of the medial temporal lobe ± grade, nor did any animals (See Implicit Cognition). generated enormously important information Cognitive science defined its subject of study ± the about the critical role of the hippocampal formation human mind ± as an `information processing' in memory (See Amnesia). The capacity of such device, and models of such things as perception, patients to learn new procedural tasks and to find memory, and the like were couched in terms of their way in new environments, without any expli- typically step-wise processes that led from an cit memory of what they were doing or why, also input through stages of transformation and repre- raised considerable interest (Scoville and Milner, sentation to a possible output (See Information 1957; Milner, 1965). Processing). A final critical discovery dependent on the study Although cognitive science at this time paid little of epileptic patients followed upon the use of cal- heed to the brain, neuroscience pushed ahead, losal section to prevent the spread of the epileptic making great strides in a number of areas. In the focus from one brain hemisphere to the other. late 1940s, the discovery of the `reticular activating These `split-brain' patients were quickly shown system' (Moruzzi and Magoun, 1949) had a major (by Roger Sperry and his colleagues)19 to have re- impact (See Reticular Activating System). This markable psychological characteristics that have landmark event shifted thinking about the brain informed us for nearly 50 years about how cogni- in a fundamental way. It showed that, contrary to tive functions are carried out in the brain. prior notions, the brain was not a passive organ Two very important meetings that were held in waiting to respond to external stimulation. Instead, the 1950s involved many of the neuroscientists it was constantly active, and the critical question involved in the work just described. Both were was no longer what brought it into activity but sponsored by the Council for International Organ- rather what kind of activity it engaged in. The izations of Medical Sciences (CIOMS). The first was xx What is Cognitive Science? held in August 1953 in the Laurentian Mountains meetings was interdisciplinary, and though these near Montreal, and brought together researchers efforts were not in the mainstream of cognitive in various fields to discuss the implications of the science at the time, they were important in setting reticular activating system ± it was titled `Brain the agenda in cognitive neuropsychology.26 Mechanisms and Consciousness' (Delafresnaye, At a more neurophysiological level, tremendous 1954).20 The second, held in August 1959 in Monte- discoveries were being made, largely in the wake video, Uruguay, brought together an even wider of technical advances that made possible the array of neuroscientists under the title `Brain Mech- recording of activity from individual neurons in anisms of Learning' (Delafresnaye, 1961).21 response to carefully controlled inputs. Here, the The inclusion of scientists from the USSR and work of Mountcastle (1957) in the somatosensory Eastern Europe at the second meeting was particu- system, and Hubel and Wiesel (1962a,b) in the larly noteworthy, as the cold war had precluded visual system, stand out as seminal, to be followed such interactions for much of the period 1946±1957. by a literal explosion of studies that continues to Once again, the Macy Foundation had an import- the present day. In this vein it is also important to ant role to play, sponsoring (with the National mention the classic study by Lettvin et al. (1959) in Science Foundation) three yearly conferences be- the visual system of the frog. These authors claimed ginning in 1958 on `The Central Nervous System to have found the neuronal correlates of Kant's and Behavior' (Brazier, 1958, 1959, 1960; also see synthetic a priori knowledge (See Kant, Immanuel), Brazier, Mary A. B.). The first meeting had a central propositions that were not derived from the exter- goal of bringing Russian neurophysiology to the nal world, but that were nonetheless taken by the West; although no scientists from the USSR were organism to be true (e.g. the axiom that a straight present, the work of Sechenov, Pavlov (See Pavlov, line is the shortest distance between two points). Ivan Petrovich), Bechterev and others was the McCulloch (1988) later referred to this study as a focus of the discussion.22 The second conference first step towards experimental epistemology. All broadened this base by including several promin- of these studies showed that the activity of neurons ent researchers from Eastern Europe ± Bures could be related in meaningful ways to certain (Prague, Czechoslovakia),23 Grastyan (Pecs, Hun- properties of external stimulation without being a gary) and Rusinov (Moscow, USSR). The third passive copy of the surrounding scene, and in so meeting included Luria (See Luria, Alexander R.) doing began the process of explaining how internal and Sokolov, both from Moscow. Yet another crit- models of the world could be instantiated in the ical meeting held at this time (October, 1958) was brain (See Neurons, Representation in). the `Moscow Colloquium on Electroencephalog- Another key finding was the discovery by Olds raphy of Higher Nervous Activity' (Jasper and and Milner (1954) of systems in the rat brain that Smirnov, 1960). This meeting ranged widely over subserve reward (See Brain Self-stimulation; Olds, many topics, and brought together the most prom- James; Reward, Brain Mechanisms of). Although it inent neuroscientists from both East and West. has taken nearly half a century for the study of The impact of these six meetings, focused on the affect to be reintegrated with the study of thought, brain, was immense. It is not an exaggeration to say reasoning and `pure' cognition, the basis for this that an entire generation of cognitive neuroscien- synthesis was laid in these early studies. This find- tists (although not yet called by that name) was ing was important at that time for another reason: it weaned on the books from these meetings. While contributed to the demise of Hull's drive reduction the meetings focused largely on arousal, memory, theory. Along with contemporary studies demon- perception and the like, similar undercurrents were strating the power of curiosity and stimulation at play in the study of language. seeking in the perceptual domain, self-stimulation By the mid 1950s, the view had emerged that the of the brain presented a form of behavior that careful study of aphasia from a variety of perspec- simply could not be accounted for in terms of tives could yield real gains in understanding lan- drives and their reduction. As the 1960s dawned, guage (See Aphasia). This feeling led to a six-week one could get the feeling of great progress, but in a seminar, held in 1958 at the Boston VA Hospital.24 compartmentalized way. The decade itself ended The discussions at this seminar were captured in with the publication of a landmark book by Miller a book published some years later (Osgood and et al. (1960) that proposed a model for cognitive Miron, 1963). A few years after the Boston VA function applying the insights of Craik (internal meeting, another meeting focused on aphasia was representations) and cybernetics (feedback) to the held in London, sponsored by the Ciba Foundation problems identified by Lashley (serial order) and (De Reuck and O'Connor, 1964).25 The spirit of both Chomsky (generative linguistics). What is Cognitive Science? xxi

CONSOLIDATING THE GAINS the linguistic world by proposing a `deep' structure that represented the essential thematic relations The 1960s were a period of consolidation of the between verbs and arguments, regardless of their gains made in the previous decade, but new prob- surface order (harking back to the `inner form' of lems were looming. Little progress was being made sentences proposed by Wilhelm Wundt and in connecting minds with brains, and the grand others). Miller and his students created a set of promises issued by proponents of the symbolic experiments that seemed to demonstrate the `psy- approach to artificial intelligence were beginning chological reality' of deep structure in memory for to look unattainable. It seemed that there were sentences. Thus was born an exciting period in quite a few things that human brains could which it seemed that results from the highly do much better, and even faster, than computers. speculative and theoretical discipline of linguistics The realization that visual recognition was actu- could be immediately tested in the laboratory: ally a very complex phenomenon followed from the rallying cry was `one linguistic rule±one mental repeated failures to get computers to solve even operation'. Extended and careful research as simple recognition problems (See Computer Vision; well as logical argument later disproved this Pattern Recognition, Statistical). Language transla- idea. Fodor and Garrett (1966) noted that the tion by machines, as we have said, also seemed a lot relationship between the grammar and behavior harder than was once imagined (See Machine must be `abstract' to some degree: Bever (1970) Translation). During these years an ongoing pro- proposed a direct `strategies' model of comprehen- gram of informal meetings was organized through sion that shortcuts linguistic rules. Their joint book the `Neurosciences Research Program' sponsored (Fodor et al., 1974) laid out systematically the by MIT and several federal agencies. Francis O. implications of these ideas for the major areas of Schmitt was the force behind this effort at the psycholinguistics: acquisition, perception and pro- outset, which brought an interdisciplinary group duction.27 of scientists together on a regular basis to discuss The fierce debate about the linguistic (or proto- the latest developments. One such meeting, for linguistic) capacities of higher primates started in example, held in 1964, was concerned with `math- earnest in these years (Gardner and Gardner, 1971; ematical concepts of central nervous system func- Gardner, 1989; Fouts, 1989), opposing the unshak- tion'. These meetings, and the reports they able believers in the evolutionary continuity of all generated that were published as a Bulletin and cognitive functions28 to the bona fide linguists, led informally circulated, had a significant impact on by Chomsky, who stressed that the unique central the emerging domain of neuroscience. components of human languages (discrete infinity, A signal event in the 1960s was the publication boundless recursivity, constituency, generative of Ulric Neisser's book, Cognitive Psychology (1967). power, compositionality) have no counterpart in This was the first textbook in the field, and it had a the communicative systems found in animals, powerful didactic and organizational influence for higher primates included (See Animal Language; many years. Sternberg's (1966) work on memory Language Acquisition by Animals). The negative stages using reaction time studies (See Reaction conclusions on the linguistic capabilities of chim- Time) and Posner's (1969) approach to abstraction panzees by the prominent cognitive primatologist and recognition made important contributions to David Premack (Premack, 1972, 1986), and the thor- this new domain. ough longitudinal study of the male chimp Nim While many areas of cognitive science were in a Chimpsky at Columbia University (Seidenberg consolidation phase in the 1960s, psycholinguistics and Petitto, 1987; Terrace et al., 1979) reset the blossomed into a highly influential discipline (See debate for quite some time. Though remarkably Psycholinguistics). The groundwork was laid in intelligent and capable of sophisticated cognitive part by an ongoing seminar at MIT on language operations, chimpanzees are provably devoid of acquisition led by Roger Brown and attended the most central components of human linguistic regularly by Chomsky and others, which formu- competence. While productive comparative studies lated the basic approach to the study of language between animals and humans in the domains of acquisition that exists today. At the same time, vision, motor control, brain development, acoustic Miller inspired a group of young psychologists to perception and categorization were destined to take as an object of experimental investigation blossom from the early 1960s to the present day, Chomsky's theory of grammar as presented in his the comparative study of language dwindled as a Syntactic Structures (1957). That model had rocked result.29,30 xxii What is Cognitive Science?

In the period beginning in the late 1960s, neuro- THE ROYAUMONT MEETING: A science had pushed forward as well, but now the DEBATE BETWEEN (AND AROUND) contributions were of a sort that clearly could con- JEAN PIAGET AND NOAM CHOMSKY, nect with cognitive science. Work on the visual 1975 system began to make contact with perception as studied in cognitive laboratories; (Blakemore and The Royaumont meeting was motivated largely by Cooper, 1970; Gross et al., 1972; Zeki, 1978), and a an attempt to `reconcile' Chomsky's approach to line of research on vision began that has yielded language with Piaget's approach to cognition in tremendous insights into how we see the world, general.32 Both Chomsky and Piaget professed a and why we see it the way we do. Much of this deep link with biology, so a reconciliation seemed research depended on neurophysiology carried out possible. Piaget opened the conference with a sum- in primates, who in the early studies were anesthe- mary of basic assumptions that he believed would tized, then only stabilized, and in recent years even be received as innocent, obvious and hardly worth capable of moving about in the world. Studies discussion. Much to his amazement, the whole carried out in ever more natural environments are meeting (three days) was dedicated to a multi- producing increasingly sophisticated understand- faceted discussion of these very assumptions. The ings of how the brain subserves vision. biologists questioned Piaget's reliance on auto- Equally dramatic was a series of discoveries regulation without specific pre-existing regula- about the hippocampus, a brain structure impli- tors33 and his attempts to reintroduce in subtle cated in memory since the pioneering work on ways the inheritance of acquired traits. Chomsky H.M. in the 1950s. O'Keefe and Dostrovsky (1971), offered basic facts about language (mostly syntax) using new methods to record the activity of single that, he claimed, could not be even remotely ex- neurons in freely moving animals, discovered plained in terms of abstractions from motor sche- `place cells' in the hippocampus, neurons whose mas, nor by any general conceptual grasp of the activity reflected the animal's location in the envir- world. Fodor argued that genuine conceptual nov- onment (See Place Cells). This discovery provided elty and any genuine potentiation of a pre-existing the basis for O'Keefe and Nadel's (1978, 1979) language could not be the result of learning. Other theory that the hippocampus instantiated `cogni- participants added their bit of specialized know- tive maps' of the sort postulated by Tolman (1948) ledge, some defending Piaget (notably his collabor- (See Animal Navigation and Cognitive Maps; Navi- ators Cellerier and Inhelder, but also Papert, gation; Navigation and Homing, Neural Basis of). Bateson, Wilden and Toulmin), others siding with This was one of the first neurophysiological re- Chomsky and Fodor (Premack, Sperber, Mehler, search programs that made a direct connection be- Piattelli-Palmarini and, in indirect ways, Monod, tween the activities of individual neurons and Jacob and Changeux). The core issues, as they complex cognitive activities. Bliss and Lomo (1973) now appear more clearly in hindsight, were: (a) discovered a form of synaptic plasticity in the the modularity of mind and the autonomy of hippocampus (long-term potentiation, or LTP) that syntax; (b) the specificity of innate cognitive struc- seemed to verify Hebb's early speculations about tures and the poverty of the stimulus; (c) reasons to how learning might occur in the nervous system keep rejecting (a) and (b) in spite of what Chomsky (See Hebb Synapses: Modeling of Neuronal Select- and Fodor presented as overwhelming evidence ivity; Long-term Potentiation, Discovery of). Re- that one should accept them. Sequels of these issues search on the hippocampus has continued at a and codas to point (c) are still very much alive as furious pace ever since (See Hippocampus). we write. While the arguments put forth by Another research program worth noting in- Chomsky and Fodor remain to many as strong as volved the efforts of a displaced psychiatrist, Eric they were then, unshakable resistance to modular- Kandel, to uncover the neural mechanisms of learn- ity, specificity and innateness survives in many ing and memory. Taking the bold step of shifting quarters of cognitive science in various incarna- his research attention to an invertebrate (the sea tions (Bates and Dick, 2000; Cowie, 1999, 2000; slug, Aplysia californica), Kandel began the process Elman et al., 1996; Karmiloff-Smith, 1992, 1994; Kar- of painstakingly working out the synaptic mechan- miloff and Karmiloff-Smith, 2001).34 In particular, isms underlying various forms of plasticity, laying debate rages over just how `impoverished' the en- the basis for an understanding of the cellular and vironment of the growing infant really is, and molecular mechanisms of memory (See Learning in whether or not powerful abilities to extract statis- Simple Organisms).31 tical regularities from the environment might not What is Cognitive Science? xxiii make possible the ontogenetic, rather than phylo- ready to be intertwined. It is safe to say that the idea genetic, acquisition of various aspects of language that the brain comprises a large number of special- including syntax. ized modules is now the accepted wisdom ± the As of the mid-1950s, Chomsky had argued for the challenges we face lie in figuring out how these `autonomy of syntax', offering more or less inciden- semi-autonomous systems interact to generate cog- tally the now famous example of the sentence `Col- nition and behavior (See Modularity; Modularity in orless green ideas sleep furiously', which every Neural Systems and Localization of Function). speaker of English identifies as syntactically well- In the 1970s and early 1980s, cognitive scientists formed even though it is utterly meaningless. made considerable strides, in particular in the Subsequent studies by Chomsky and his early col- study of spontaneous mental imagery (later assem- laborators (Chomsky, 1965) revealed basic syntactic bled and expanded in Kosslyn, 1980; also see Im- principles of a very specific nature, common to agery); mental rotation (Shepard, 1971; also see many, and arguably to all, languages and dialects, Mental Rotation); concept and category formation as an integral part of the speaker±hearer's tacit (Rosch, 1973, 1978; Smith and Medin, 1981; also see `knowledge of language'. These did not resemble Concept Learning; Concept Learning and Categor- in the least the then known basic principles of ization: Models); biases and heuristics in natural visual perception, motor control and generic reasoning and decision making (Kahneman et al., reasoning, forming an integrated cluster of autono- 1982; Kahneman and Tversky, 1972, 1973, 1979; also mous cognitive rules and representations. `Know- see Reasoning); memory (Tulving and Thompson, ledge of language' had to be kept separate from 1973; also see Memory; Memory Models; Memory: generic knowledge of the world. The metaphor of Implicit versus Explicit); abstraction (Posner, 1978); a `language organ' defied all traditional concep- motion perception (Johansson, 1973); cognitive tions of a small set of `horizontal' multi-purpose conceptual development in the child (Carey, 1985; mental faculties. These two separate strands of a Keil, 1979; Markman, 1989; Spelke, 1985, 1988; also modular conception of the mind±brain, one focus- see Categorization: Development of);35 learnability ing on language and `input systems' (Fodor, 1981, theory (Osherson et al., 1986; also see Learnability 1983) at the level of mental contents, representa- Theory); and the theory of Government and Bind- tions and symbols, the other on central systems ing (Chomsky, 1981; also see Binding Theory; (memory, perception and planning) at the level of Government±Binding Theory). neuronal substrates in the animal and in humans, A major paradigm shift was in the offing, how- were to converge eventually, though emanating ever, in the revival of biologically-inspired ap- from different starting points. proaches to cognitive science. Beginning with the On the neurobiological front, in the early 1970s efforts of a group in San Diego, the `connectionist' Tulving (1972) suggested that there might be two movement has made major inroads in a large rather different kinds of human memory ± episodic number of fields previously dominated by the and semantic. In 1974, three papers were published views first emphasized by Newell and Simon (See suggesting the same thing, but based instead on Connectionism). A critical focus of this approach animal research (Hirsh, 1974; Gaffan, 1974; Nadel was its assertion, sometimes explicit, sometimes and O'Keefe, 1974). This notion applied to the brain implicit, that cognitive models should look more an idea first promulgated by Tolman in a classic closely at biology. Instead of emphasizing the sym- paper, `There is more than one kind of learning' bolic level that had been the bread and butter of (1949). It was discussed by O'Keefe and Nadel cognitive science since the mid-1950s, this ap- (1978) for both animals and humans as part of proach claimed to eschew symbols altogether, their `cognitive map' theory, and was also applied focusing instead on distributed representations within the human amnesia literature by Kinsbourne and learning algorithms (Rumelhart et al., 1985; and Wood (1975) and then Cohen and Squire (1980). McClelland et al., 1985; also see Distributed Repre- The general notion that there are multiple neural sentations). Connectionism as a concept is an old modules concerned with different kinds of memory idea ± Hebb (1949) referred to it in his book. The is now widely accepted (Schacter and Tulving, label resurfaced in the mid-1980s as the name for a 1974). A somewhat similar history unfolded in the new approach to neural networks, one that has had study of visual cognition, beginning with the sem- a major impact on the domain of cognitive science inal work of Ungerleider and Mishkin (1982) on in the past 20 years. Initially developed by John J. `two visual systems'. When a few years later Jerry Hopfield in 1982, neural networks were abstract Fodor published his landmark philosophical trea- entities (later to be also implemented by physical tise The Modularity of Mind (1983), the strands were hardware) that were explicitly inspired by real xxiv What is Cognitive Science? neuronal circuitry and were capable of automatic called `principles and parameters' (PP for short). In learning, rule extraction and generalization (Hop- essence, this model reduces all the differences be- field, 1982). Hopfield showed how the mathemat- tween human languages to a small universal set of ical simplification of a neuron could allow an syntactic nodes (parameters), for each of which analysis of the behavior of large-scale neural net- there is a choice between only two admissible works, modeling progressive descents on an `values'. The binary values for each parameter are energy surface, thereby mimicking automated labeled as `+' (the marked value) and `±' (the un- learning and automatic feature extraction from a marked or default value). Under this idealization, corpus of different, but related, stimuli. Under an the `task' of the child learning his/her native lan- assortment of training procedures, with artificial guage consists in appropriately `fixing' the binary equivalents of reinforcement and punishment, values of all the parameters in conformity with the and with `backpropagation' from one layer of ter- set of values implicitly chosen by the surrounding minal nodes back to layers of `hidden' nodes (a community of speakers. James Higginbotham has powerful improvement that could solve some of summarized this idealization as the positioning the problems of older perceptrons), the remarkable of a set of `switches' on a mental `panel' (Higgin- potential of these artificial networks created a sen- botham, 1982).37 sation (See Backpropagation). Cognitive scientists, in a number of places, paid very close attention and THE EMERGENCE OF COGNITIVE began to challenge the modularist±innatist theory NEUROSCIENCE: SOME EXCITING of mind, and the very idea of dedicated mental DEVELOPMENTS rules and representations. The efficiency of such connectionist networks in extracting common fea- Progress in cognitive neuroscience during the last tures from certain families of inputs could equal, or decades has been nothing short of phenomenal, even surpass, that of humans, as Stephen Gross- owing in large measure to the development of berg of Boston University had noticed some years neuroimaging techniques that have made it pos- before (Grossberg, 1976; also see Adaptive Reson- sible to study the human brain during various cog- ance Theory). But in spite of the adjective `neural' nitive activities (See Neuroimaging). The use of and in spite of the liberal use of terms borrowed electroencephalographic methods with humans from biology (evolutionary landscape, fitness, has a long history, but such methods involving adaptive behavior, etc.), the real proximity to surface recordings have inherent limitations, most `wet' neurobiology remained questionable.36 specifically related to the spatial localization of the The rise of connectionism in cognitive science recorded signals. The use of many more recording goes hand in hand with neo-Piagetianism, touting sites and sophisticated analytic tools has engen- the virtues of general intelligence and multi-pur- dered a new generation of such methods, most pose cognitive mechanisms powered by processes prominent in the domain of event-related poten- called stepwise abstractions, categorizations, them- tials (ERPs), where recordings are synchronized to atizations and generalizations. Renewed invitations cognitive events of interest so that patterns of brain to go `beyond' innatism and modularity (Bates and activation specifically related to those events can be Elman, 1996; Elman, 1989; Elman et al., 1996; Kar- identified. Such ERP methods have yielded consid- miloff-Smith, 1992, 1993, 1994) show how pertinent erable insights, especially in the temporal domain. the arguments and counterarguments developed at However, spatial localization remains a problem. the Royaumont conference remain. The often feisty This is where the emergence of new neuroimaging debates between proponents of these two ap- techniques has been most productive. The critical proaches to cognitive science continue to this day. insight here was the realization that methods could Since the early 1980s, a conception of learning be devised to track metabolic and other conse- that is radically different not only from the connec- quences of neural activity in humans. The first tionist models but also from almost all previous widely used method, positron emission tomog- models of learning has been developed in a new raphy (PET), depends upon the use of radioactive approach to the study of first-language acquisition substances and the uptake of these materials by (notice that even the word `learning' has been ex- recently active neural tissue. More recently, a less punged). Commandeering and reorganizing an invasive method, functional magnetic resonance array of studies on many languages and dialects imaging (fMRI), has been developed as an alter- in a series of lectures at the Scuola Normale in Pisa native. This method takes advantage of the fact (Italy) in 1980, and then in a published book that blood oxygenation levels change as a function (Chomsky, 1981), Chomsky introduced the model of neural activity, and that oxygenated and What is Cognitive Science? xxv deoxygenated blood (or, more precisely, hemo- Galese, 1998; also see Simulation Theory). The pos- globin) have different magnetic properties. This sible role of a system of mirror neurons in the permits the detection, with powerful magnets, of creation of internal mental models is obvious, and those areas of the brain mobilized by some form of the implications of these findings for theories of the cognitive activity. Yet another method, magneto- emergence of language is under active discussion encephalography (MEG), depends upon the very (e.g. Rizzolatti and Arbib, 1998, 1999). Another small, but measurable, magnetic fields engendered major advance concerns the development of by neural activity. This method, though depending methods for simultaneously recording from many on considerable analysis to extract signals from individual neurons, making it possible to study the noise, offers great promise given its ability to activity of neural ensembles. These methods have couple the real-time dynamic response (on a scale been quite productively applied to the study of of milliseconds) with accurate spatial localization. hippocampal `place cells' (e.g. Wilson and Finally, transcranial stimulation (TMS) has emerged McNaughton, 1993), where it has been possible to as a method to stimulate or, mostly, selectively in- demonstrate that patterns of activity observed hibit areas on the cortical surface, and has been during a rat's daily activity have a high likelihood productively used to study in a very precise way of recurring during a subsequent sleep episode. the role of these surface structures in various cogni- New findings from developmental neuroscience tive functions. and neuroanatomy have overturned the long- Considered together, these methods have accepted view that nerve cells cannot be formed brought about a considerable explosion of research after the earliest stages of life (e.g. Gould and on the brain mechanisms of normal human cogni- Gross, 2002; also see Neurogenesis). This, and tive function.38 Where previous studies were other findings from the study of memory and per- limited to pathological cases and involved the ception, have reminded neuroscientists of the in- problematic analysis of function from the deficits credible dynamism of the brain. A major challenge caused by pathology, these methods provide an for the future, at the very heart of the cognitive entirely new window into the human mind. Not science enterprise, is to figure out how the stable surprisingly, they have in the first instance con- world our minds construct, as pointed out by firmed many of our hard-won assumptions about Craik, can be instantiated in a biological substrate which parts of the brain are engaged by what kinds that is constantly changing. Or, to put it as McCul- of cognitive activity. Imaging studies have also loch (and Shakespeare) did: `What's in a brain that shifted the focus of explanation away from reliance ink might character' (McCulloch, 1964). on discrete `centers' of cognitive function towards Thanks to the stunning discovery of systematic the notion of an interaction between multiple brain errors in spontaneous reasoning and decision areas. While some might argue, even at this stage, making, notably by Amos Tversky and Daniel that neuroimaging is merely a modern-day version Kahneman, a progressive integration has begun of phrenology (activations in the head, rather than between cognitive science and economics (for a bumps on the skull; Bates and Dick, 2000), clever pioneering survey, see Thaler, 1991, 1992). The re- researchers are beginning to develop new para- cently explored neuronal bases of decision making, digms that offer the promise of real advances that both in pathological cases (Adolphs et al., 1996; would have been impossible with earlier tech- Bechara et al., 1994, 1999; Damasio, 1994; Damasio niques. Finally, the combination of neuroimaging et al., 1996) and in normal subjects (Breiter et al., with more traditional single and multiple neuron 2001) has suggested that a whole new domain, recording and with selective chemical labeling called neuroeconomics (McCabe et al., 2001; Smith, methods offers the promise of combining the in- in press; also see `Neuroeconomics) may be just sights that these approaches can offer.39 around the corner. The need to integrate standard The productive use of more traditional methods economic analyses with what cognitive scientists has by no means ceased: consider for example have discovered about spontaneous heuristics and the discovery in monkeys of the `mirror neurons' biases is now reported in the popular press.40 (Rizzolatti and Craighero, 1998; also see Mirror Neurons). These neurons demonstrate a remark- THE POSTMODERN ERA able property: the same neuron is active when the animal either engages in an action or observes an- Among the most striking changes in cognitive sci- other animal engaging in the same action. The ex- ence since the 1980s years has been the shift in what istence of such neurons raises questions of great is open to study. As we noted earlier, cognitive import to philosophers of mind (Goldman and science started with the view that cognition is xxvi What is Cognitive Science? limited to those things humans can be conscious et al., 2001; Ramus, in press; Ramus et al., 1999; of. This position has been totally abandoned, and also see Language Acquisition). much of the domain is now concerned with phe- Critics of modularity are also to be found in nomena that lie behind the veil of consciousness. cognitive neuroscience, notably in the analysis of Whether or not they are conscious, animals are very pathological deficits heretofore depicted as para- much a part of modern-day cognitive science. One digms of modularity ± prosopagnosia, for instance, prominent example concerns the study of emotion. with contrasting views, and contrasting observa- Great strides have been made in linking emotion to tions, championed by Isabel Gauthier (Gauthier traditional views of cognition, thereby returning et al., 1999, 2000a,b) on one side, and by Nancy the field to the point at which Craik left it more Kanwisher and Morris Moscovitch (Kanwisher than 50 years ago. Finally, the grand-daddy of them and Moscovitch, 2000) on the other (See Prosopag- all, consciousness itself, has become the focus of nosia). The nature and significance of earlier dis- intense research interest within cognitive science coveries of specific language variants and deficits, (and beyond) in recent years (See Consciousness).41 from savants (Smith and Tsimpli, 1995) to Williams Cognitive science has now reached the stage syndrome (Bellugi et al., 1994, 1999, 2001; Bellugi where one sees the production of integrated text- and St. George, 2000; Stevens and Karmiloff-Smith, books (Osherson, 1990/1995; Stillings et al., 1987; 1997; also see Williams Syndrome); to sign- Bechtel and Graham, 1999; Posner, 1989) and the languages (Kegl and McWorther, in press; Kegl publication of a concise encyclopedia (Wilson et al., 2000; Klima and Bellugi, 1979; Petitto, 1987; and Keil, 2000). Neuropsychology and cognitive Petitto and Marentette, 1991; Emmorey and Lane, neuroscience also now have their comprehensive 2000; also see Sign Language), to SLI (Specific Lan- sourcebooks (Posner, 2001; Gazzaniga, 1984, 2000; guage Impairment; Gopnik, 1990, 1994; Van der Kosslyn and Andersen, 1992). Lely, 1997; Van der Lely and Stollwerck, 1996; Wex- Having grown into a rich and multifaceted ler, 2002), is being questioned by researchers who domain, it is no surprise that cognitive science has conceptualize language as a specialization of gen- witnessed, is witnessing, and will continue to wit- eral cognitive and communicative functions (Bates ness disagreements, schisms, partial reconciliations and Dick, 2000; Bates et al., 1999; Karmiloff and and yet further splits in theories and methodo- Karmiloff-Smith, 2001; Karmiloff-Smith, 1998; Vol- logical criteria. If we decide, with some simplifica- terra et al., 2002; Volterra and Erting, 2002). Even tion, to characterize as `mainstream' or `classical' the legitimacy of combining data from language cognitive science the individualist, largely innatist, pathologies in the adult with data on developmen- modular and representational±mentalist (RTM) tal deficits in the child is being criticized in conception of the mind that characterized much of principle (Karmiloff-Smith et al., in press). the 1980s and 1990s, there are clear signs that we Theories of language evolution that revise the may be entering a postclassical cognitive science approach of generative grammar (Pinker, 1994; (Piattelli-Palmarini, 2001). The innovative turn Pinker and Bloom, 1990)42, only pay lip service to introduced by connectionist models in the mid- it (Deacon, 1997), or fly in the face of decades of 1980s has revamped an anti-modular and general- research in generative grammar (Dunbar, 1999; Lie- purpose conception of the mind±brain, soon berman, 2000; Tomasello, 1999), have recently been contested by `classic' cognitive scientists (Pinker published. The age-old attempt to derive linguistic and Mehler, 1988). The last few years have wit- structures from motor control, considered mori- nessed a partial (only partial, but not irrelevant) bund (cf. the exchange between Chomsky and Pia- rapprochement between the two camps: many get on this point at the Royaumont debate described connectionists now countenance initial sets of pre- earlier), is being revamped under a different guise wired connections, and can explain the spontan- (Rizzolatti and Arbib, 1998, 1999). Cognitive innat- eous tendency of parallel networks to locally ism is being re-analyzed at its roots, and allegedly cluster into modules (these modules are, however, more promising alternatives are being offered conceived as emergent aÁ la Piaget, not prewired; (Cowie, 1999, 2000; Elman et al., 1996). Elman et al., 1996; Karmiloff-Smith, 1992, 1994). The cognitive sciences today expand in every Symmetrically, on the other front, several linguists direction, as can be seen in the wide range of art- and developmental psychologists rooted in the icles included in this encyclopedia. Neuroscience generative tradition are presently searching for in- has been drawn back into the fold, and the area of ductive (even statistical) components of early lan- cognitive neuroscience is one of the major growth guage acquisition, and report finding some that industries in the field. We will let the articles in this may play a crucial role (Nespor, 2001; Newport work speak for themselves in filling in our history. What is Cognitive Science? xxvii

Among these articles are biographies of a variety of essay from near oblivion and justified the subsequent pioneers (now deceased) whose contributions were characterization of Freud, by Frank Sulloway, as a critical to the emergence of cognitive science. Also `biologist of the mind' (Sulloway, 1979). However, among these articles are a number of overarching Freud's approach went gloriously off the rails, and reviews that attempt to address large domains was ultimately abandoned by Freud altogether. This failure can now be seen as inevitable, given the tools within cognitive science (See Action; Cognitive Sci- that Freud had to work with at the time. ence: Experimental Methods; Cognitive Science: 6. Somewhat later, spelling out in full a judgment that Philosophical Issues; Consciousness; Development; had been implicit all along, the pro-behaviorist ana- Perception). lytic philosopher Quine referred to these mind- It is impossible to anticipate the paths that cogni- internal entities as `creatures of darkness' (Quine, tive science is going to take in the years to come. We 1956; See Behaviorism, Philosophical). It was inevit- are entering a postclassical era and there are able that the chief architects of an unrepentantly reasons to believe that it will prove as productive mentalistic, internalist, computational and representa- and as innovative as the one that preceded it. This tional theory of mind (usually referred to as RTM ± encyclopedia offers a complete and complex pic- Representational Theory of Mind), most notably ture of the discipline up to the present. What Noam Chomsky and Jerry Fodor, found themselves fighting a long and sustained battle against be- will happen in the future will almost certainly haviorism and its avatar in contemporary analytic surprise us. philosophy (methodological behaviorism). Chomsky's Putting together the encyclopedia has been a destructive review of Skinner (Chomsky, 1959/1980) challenge but also an opportunity, and the same and Fodor's anti-behaviorist book-long essay The Lan- can be said about looking into the history of the guage of Thought (Fodor, 1975) helped shape the turn field. It is a fascinating history, peopled by intellec- away from behaviorism and much of modern mental- tual giants and featuring ruminations about the big istic cognitive science for many years. Chomsky issues that have concerned thinkers for several mil- argued in favor of the specificity and universality of lennia. A complete history of these times remains to grammar. He defended an internalist, individualistic, be written; perhaps a reader of this encyclopedia intensional (as opposed to extensional) characteriza- will be sufficiently excited by its contents, and the tion of grammars (Chomsky, 1955, 1956, 1957, 1986) against a Wittgensteinian conception of language as a genesis of these ideas, to take on that task. collective conventional public entity, individually mastered through the tuning up of a `skill', eternally NOTES subject to the cumulative action of infinitesimally small variations, from one dialect to the next, from 1. We gratefully acknowledge the help of several col- one generation to the next. (cf. Chomsky's and Fodor's leagues, Tom Bever, Howard Gardner and Morris writings in reaction to theses by Quine, Putnam, Moscovitch, who read an early version of this docu- Davidson, Dummett, Kripke: Chomsky, 1980, 1988, ment and whose suggestions vastly improved it. They, 1995, 1998; Fodor, 1981, 1990; Fodor and Lepore, of course, are not responsible for its remaining short- 1992). comings. 7. Hull's attempt to generate mathematical treatments of 2. A century later, matters are not quite as settled as they learning, especially animal learning, was brought into appeared to be. There is now considerable evidence of the modern era by Rescorla and Wagner (1972) and direct contacts between at least some neurons in the Mackintosh (1974). Reacting to critical new findings CNS ± so-called electrical synapses. such as the phenomenon of `blocking' (e.g. Kamin, 3. In 1906, Golgi and Cajal were jointly awarded the 1969), they produced learning rule equations that an- Nobel Prize `in Physiology or Medicine'. ticipated some of the more powerful algorithms to 4. There were important `schools' of neuropsychologists be developed within the connectionist framework and neuropsychiatrists throughout Europe at this 20 years later. (See Animal Learning) time. In Germany, Brodmann, Pick, Alzheimer, and 8. Brunswik influenced Rosenblatt, who developed Korsakoff are noteworthy. In France there were Janet the perceptron model in the 1950s. He also had a and Charcot, among others. In England, Hughlings- significant influence on Julian Hochberg, who two Jackson, Head and Ferrier deserve note. In Russia, decades after his student days at Berkeley talked Sechenov was extremely influential, and had a major about perception in terms of piecemeal perception, impact on Pavlov. mental structures and the intentions of the viewer. 5. An interesting example is offered by Freud's Project for Hochberg emphasized integration over inputs a Scientific Psychology (1895). This remarkable piece of obtained from successive glances, hence the critical work started out in a blaze of glory, defining in a very role of memory and other cognitive factors in percep- clear way what the issues were, and what kinds of tion, a line that has been continued by Hochberg's answers would be necessary. Indeed, Karl Pribram students to the present day (e.g. Peterson, 1999; also (Pribram and Gill, 1976) resurrected this remarkable see Vision: Top-down Effects). An explicit revival of xxviii What is Cognitive Science?

Brunswikian models in decision making is presently interspersed with `noise', in conformity with the then advocated by Gerd Gigerenzer and his colleagues at influential theory of information. the Max Planck Institute for Human Development in 12. According to Hebb, perception could be accounted Berlin (Gigerenzer et al., 1991); by A. J. Figueredo at for in terms of organized sets of neurons (cell as- the University of Arizona (Figueredo, 1992); and by semblies); thought would then follow from the con- Kenneth R. Hammond, now Emeritus Professor at catenation of many of these assemblies into phase the University of Colorado in Boulder, who has re- sequences. Memory involved changes in the efficacy cently edited an extensive collection of reprints of of connections between neurons composing these Brunswik's original papers (Hammond and Stewart, ensembles. This deceptively simple approach had a 2001). dramatic impact on that subset of investigators 9. Piaget's work covered a variety of crucial domains, willing to pay attention to the brain at the time, and from the child's conception of numbers to moral over the past few decades has had an even wider judgment, from the development of the concept of effect. In Hebb's time his theory led to a variety of causality to a cognitive approach to the historical important research findings, including work on the development of physics and of science in general effects of sensory deprivation, the stabilization of (See Piagetian Theory, Development of Conceptual visual inputs to the retina, the impact of enriched or Structure in). Many schoolteachers and avant-garde deprived rearing conditions, and a good deal more. research teams in developmental psychology had 13. This meeting was presided over by Henry Brosin, a tried with some success to apply Piaget's central psychiatrist from Pittsburgh who played a significant ideas to their teaching in the classroom. The progres- facilitatory role throughout this early era. Brosin later sive spontaneous unfolding of higher cognitive cap- retired to Tucson, Arizona, and a fair number of his acities in the developing child had been modeled by books from these formative days, copiously anno- Piaget and his collaborators at the University of tated, have found their way into the hands of one of Geneva through a series of successive `stages', char- us (L.N.). acterized as `logically necessary' and universal, that 14. John McCarthy, Marvin Minsky, Nat Rochester, the child attains one after the other at characteristic Claude Shannon, Oliver Selfridge, Herbert Gelernter, ages. Each stage was described in detail by means of Alan Newell (See Newell, Alan) and Herb Simon (See novel characteristic mental operations that were Simon, Herbert A.), among others. either absent or only embryonic in the preceding 15. The list includes, in alphabetical order: Ursula Bel- stages. The internal mental engines of this stepwise lugi, Tom Bever, Roger Brown, Jerome Bruner, Susan process had been identified by Piaget as consisting of Carey, Noam Chomsky, Jerry Fodor, Merrill Garrett, increasing auto-regulation, thematization, grouping, Janellen Huttenlocher, Roman Jakobson (See Jakob- and `reflective abstraction' (in French: abstraction son, Roman), Dan Kahneman, Jerrold Katz, Paul reÂfleÂchissante). He pictured these processes as pre- Kolers, Pim Levelt, David McNeill, Jacques Mehler, sent already, at lower levels, in all living beings, George Miller, Don Norman, Eleanor Rosch, Dan making cognitive science continuous with biology. Slobin, Amos Tversky (See Tversky, Amos), Peter It hardly needs to be stressed how radically opposed Wason, Nancy Waugh. to behaviorism this whole conception was, and how 16. Chomsky's position vis-aÁ-vis the neurobiological refreshing it appeared to many psychologists in the foundations of language and mind deserves a word 1950s and 1960s. (see text and notes 31 and 33.) of clarification. Over many years he has insisted that 10. These classical papers have been reprinted in van all kinds of relevant data (the qualification `relevant' Heijenoort (1967) and in Davis (1965). For a modern is essential here) from any domain of science or even systematic treatment see Lewis and Papadimitriou from everyday observation are of interest. Having (1981); Rogers (1988). For the impact of these theories constantly used the hyphenated expression brain± on early linguistic models see Bar-Hillel (1953a,b); mind, and having always insisted that linguistics is Barton et al. (1987); Chomsky (1956). part of the natural sciences (verbatim: part of biology 11. The intellectual climate at the Research Laboratory of `at a suitably abstract level'), he is clearly aware of the Electronics (RLE) of MIT in the early 1950s was one of potential power of the neurosciences to corroborate excitement and imminent accomplishment. The in- or refute abstract linguistic hypotheses. While en- jection of ideas and models from cybernetics and couraging a serious search for neuronal correlates, information theory into the study of language Chomsky maintains, however, that the neurosciences appeared capable of leading towards a full under- must cooperate with other relevant disciplines (e.g. standing of complex communication in humans, linguistics, developmental psychology, the study of animals and machines. The modern scientific theory first and second language acquisition, genetics) and of language, and its momentous impact on cognitive also must not forget the power of logic, of abstract science as a whole, largely originated in those la- arguments and even of physics and mathematics. His boratories, in the mid- and later 1950s, thanks to a early participation in meetings with neurologists disillusionment with both traditional classificatory (especially on aphasia and other language deficits, linguistics and the statistical approaches to lan- as we have stated here), his close interaction with guage that modeled it as a largely repetitive `signal' Eric Lenneberg, and his co-organization of a group What is Cognitive Science? xxix

focused on bio-linguistics at MIT with the molecular 20. Papers were presented by Magoun, Moruzzi, Pen- biologist Salvador E. Luria and the neuropsycholo- field, Hebb, Lashley and Kubie, among others. gist Hans-Lukas Teuber (See Teuber, Hans-Lukas) 21. Papers at this meeting were presented by Hebb, Olds, testify to his active interests in biology (for an insight- Magoun, Morrell, Hernandez-Peon, and a number of ful reconstruction of that initiative, see Jenkins others. An important aspect of this meeting was the (2000)). Some may interpret his attitude towards the inclusion of a number of key researchers from Russia brain sciences as tepid (at best) because of the quint- and both Western and Eastern Europe. Anokhin, essentially internalist±mentalist character of his the- Asratyan, Eibl-Eibesfeldt, Fessard, Grastyan, Jouvet, ories in linguistics and a consistent refusal to give a Konorski, Lissak, Naquet and Thorpe participated. privileged scientific status to data on brain structures 22. Participants included Magoun, Brazier, Doty, Olds, and mechanisms, as compared to data on linguistic Pribram, Purpura, Galambos, John, Morrell, Sperry, intuitions. Chomsky concurs with Fodor in caution- and Teuber. ing against `the intimidation by white blouses' (an 23. With whom L.N. subsequently was a postdoctoral expression used by Fodor in public discussions to fellow (1967±1970). The Prague laboratory of Jan refer to brain scientists) who present as truly scien- Bures and his wife and scientific partner Olga Bur- tific only `wet' data, as opposed to cogent and ration- esova became a mecca for neuroscientists from ally supported theories. around the world, and remains so today. Their 17. In recent times, Putnam (the acknowledged father of work on memory and more recently the spatial func- philosophical functionalism) has retreated from his tions of the hippocampus has been influential for earlier position, advocating a pivotal role for our more than four decades. intuitions about the material nature of cognitive 24. Participants included Roger Brown, Noam Chomsky, systems (Putnam, 1987, 1988). Ned Block, a former Norman Geschwind, Kurt Goldstein, Harold Good- student of Putnam's, has pointed out some serious glass, Eric Lenneberg, Brenda Milner, Charles `troubles with functionalism', requiring a consider- Osgood, Karl Pribram, and Hans-Lukas Teuber. able expansion of this conception (Block, 1978). John 25. Participants at this meeting included Macdonald Searle, over many years, has challenged the very Critchley, Lord Brain, Roman Jakobson, Donald consistency of functionalism, pleading for the cen- Broadbent, Alexander Romanovich Luria, Brenda trality of a specific causal role attributed to the Milner, Henri Hecaen, Oliver Zangwill, Hans-Lukas unique biological structure of the brain (Searle, Teuber and Colin Cherry. There were a number of 1980a,b, 1992, 1996; also see Chinese Room Argu- other important meetings held in the United King- ment, The). At the other extreme, John Haugeland dom in the 1950s and 1960s, including a series of has challenged the legitimacy of functionalism on meetings held in London in the 1950s under the title more abstract grounds, espousing the holist doctrines `London Symposium on Information Theory'. At of human cognition proposed by the German phe- the Fourth Symposium (Cherry, 1961), held in nomenologists (Haugeland, 1981, 1997). As we 1960, Averbach and Sperling reported on their write, nonetheless, some variant of functionalism methodology involving brief visual exposures to still appears to constitute the spontaneous (and subjects who were asked to report only a subset of often implicit) philosophy of the mind/brain for the presented material. Using this partial report most cognitive scientists. However, the recent devel- method, the authors were able to show that immedi- opment of refined brain imagery techniques and the ate visual memory has available to it a good deal growing number of publications dealing with specific more information than subjects can retrieve when brain correlates of higher cognitive functions (from asked for a full report. The `Mechanisation of language to numerical cognition, from decision Thought Processes' symposium held at the National making to categorization), might eventually modify Physical Laboratory in 1958 included presentations somewhat the functional conceptual scheme of cog- by Minsky, Mackay, McCarthy, Ashby, Uttley, nitive science. A direct match between abstract cogni- Rosenblatt, Selfridge, McCulloch, Sutherland, Greg- tive characterizations and real brain structures is now ory, and Bar-Hillel. Other noteworthy attendees in- increasingly possible. Except for a minority of unre- cluded Bartlett, Buerle, Cherry, Gabor, Wason, and pentant symbolists, and, at the other extreme, of irre- J. Z. Young. ducible reductionists, the complex expression brain± 26. At about this time a small group of neuropsycholo- mind is taking on a more insightful and richer unified gists, the International Neuropsychological Sym- meaning. It is interesting to realize that the core of this posium, started to meet every year in Europe. problem had already been identified, and vibrantly According to Boller (1997), the group was launched debated, in the early conferences on cybernetics. in 1949 at a party that Henry Hecaen held at his home 18. More recently, the special epistemological import- on the occasion of the International Congress of ance of the spontaneous oscillatory activities of the Psychiatry. After dinner, he outlined a proposal to brain has been stressed by Rodolfo Llinas and Jean- found an international group to promote knowledge Pierre Changeux (2002) (See Neural Oscillations). and understanding of brain functions and cognate 19. Awarded the Nobel Prize in `Physiology or Medi- issues on the borderland of neurology, psychology cine' in 1981, with David Hubel and Torsten Wiesel. and psychiatry. This group, which continues to meet xxx What is Cognitive Science?

yearly, strongly promoted the integrative, multidis- between unicellular organisms (microsemiotics), ciplinary perspective that became characteristic of plants (phytosemiotics) and the circuits of neuro- cognitive science. Later on, in Italy, cognitive neuro- transmitters in the nervous system (neurosemiotics). psychology was to flourish beyond any other domain These incremental steps in the quality and complex- of cognitive science, gaining considerable inter- ity of signaling were analyzed as accruing to a national recognition. Individuals and groups were common semiotic substrate, displaying a universal scattered in many different universities (Eduardo `perfusion of signs' which, according to Sebeok, Bisiach in Turin, Anna Basso in Milan, Carlo UmiltaÁ, authorizes a unified conceptualization, a semiotic Remo Job and Renzo de Renzi in Padua, Elisabetta `ecumenicalism' (Sebeok, 1977). Sebeok's conceptual- Ladavas in Bologna, Gabriele Miceli in Rome, to ization and his alleged semiotic `theorems' and name a few). Ever since the early 1980s, the annual `lemmas' have found attentive ears in some literary international conferences held in Bressanone in Janu- quarters and in some schools of communication ary, under the auspices of the University of Padua, (notably in Italy), but have remained, in the main, have regularly assembled the Italian `contingent' of alien to cognitive science. The semantics of natural cognitive neuroscientists with colleagues from many language has developed a radically different ap- other countries. proach (for a textbook synthesis, see Larson and 27. For a recent revisitation of the origins of psycholin- Segal (1995)). guistics and for a development of the `strategic' ap- 31. See Kandel (1980) for a review of some of this re- proach to language comprehension, see Townsend search program, begun in the 1960s, presented at a (2001). symposium held in Texas in 1978 to commemorate 28. For example, Allen and Beatrice Gardner, Duane and the thirtieth anniversary of the Hixon Symposium Sue Savage-Rumbaugh, Roger and Diane Fouts and, and to honor Lloyd Jeffress, who edited the volume from a distant shore, Jean Piaget. (For a recent re- from that earlier historic meeting. Kandel was appraisal, see Savage-Rumbaugh, in press.) awarded the Nobel Prize in 2000, with Arvid Carls- 29. Except for a few unrepentant `continuists', main- son and Paul Greengard. His monumental textbook stream cognitive science in the domain of language Principles of Neural Science, updated and translated followed different paths, unearthing deep and hard- into many languages, has been adopted in many to-detect similarities among distant languages, countries. broaching the gap between syntax and semantics 32. The published volume (Piattelli-Palmarini, 1980), (See Semantics and Cognition), exploring subtle and the afterthoughts by its principal organizer lexical structures (See Lexicon), modeling the `lo- (Piattelli-Palmarini, 1994) spare us from having to gical' problem of language acquisition with formal- summarize this rich multidisciplinary conference ized mathematical tools, developing computational (organized by Monod, Piattelli-Palmarini, Atran models of linguistic competence and performance. and Changeux). Besides Piaget and Chomsky, the The very issue of the biological evolution of language main participants from cognitive science were Jerry was to be tackled afresh at its roots by generative Fodor, Barbel Inhelder, Guy Cellerier, David linguists, steering a course away from simplistic Premack, Seymour Papert, Gregory Bateson, Dan adaptationism and continuism, examining the pos- Sperber, Scott Atran and Jacques Mehler. Other dis- sible conditions of the evolution of the very roots of ciplines were also represented: in attendance were linguistic competence (recursiveness, constituency, the molecular biologists Jacques Monod and FrancËois compositionality, infinite discreteness, generativity; Jacob, the neurobiologist Jean-Pierre Changeux, the Jenkins, 2000; Lightfoot, 1982; Nowak et al., 2002; philosopher of science Stephen Toulmin, the anthro- Piattelli-Palmarini, 1989; Uriagereka, 1998). For a pologist Claude LeÂvy-Strauss, the ethologist Norbert recent reappraisal by Chomsky himself, see Hauser Bischoff, and the mathematician Jean Petitot. Import- et al. (in press). ant additional contributions to the volume by invited 30. A special position in this debate between continuists participants who had been unable to attend came and modularist±innatists was occupied by the influ- from the logician and philosopher Hilary Putnam ential biosemiotician Thomas Sebeok. He rejected and the mathematician Rene Thom. The post-confer- wholesale all the experiments on the alleged linguis- ence exchange between Putnam, Chomsky and tic abilities of apes, claiming a much deeper, more Fodor was included by Ned Block in his anthology universal and more meaningful underlying sub- of writings, The Philosophy of Psychology. In the pref- strate: the `semiotic function'. He described incre- ace to this book, Howard Gardner suggested that the mental steps of complexification in this universal meeting was one of the seminal events at the very underlying substratum and insisted that a unified origins of cognitive science. theory could range from the `syntactic' (sic) nature 33. FrancËois Jacob (the co-discoverer, with Jacques of Mendeleeff's table of the chemical elements Monod, of the `operon', a complex genetic unit of (Sebeok, 1995/2000) up to all systems of human com- regulatory genes) stated that regulation can only take munication, be they vocal, gestural, graphic or pic- place as a result of pre-existing regulatory genes that torial, passing through the genetic code, the immune actually and selectively kick in (or remain shut off) to code, the systems of communication between cells, regulate metabolic pathways. Piaget's conception of What is Cognitive Science? xxxi

progressive cascades of higher and higher auto-regu- by an element with a magnetic dipole ± spin ± that lations was at odds with this very concrete finding. It interacts mostly, but not exclusively, with its imme- quickly appeared to some participants that Piaget's diately adjacent neighbors, with a statistical tendency concept was a metaphor, not a model. Piaget immedi- to propagate local dipole alignments to larger ately retorted that Jacob's idea was exceedingly regions, in search of global minima (attractors) in `narrow', and that he knew other biologists who con- the resulting energy surface; for a comprehensive curred with him in presenting a more general and synthesis, see MeÂzard et al., 1987). Inspired by more flexible picture of auto-regulation. models of ferromagnetic lattices developed in the 34. Over the years, readers of different scientific orienta- 1920s by the German physicist Ernest Ising (Brush, tions have drawn strikingly different conclusions 1967) and temperature-sensitive probability decay from the proceedings of this debate. One of us functions initially proposed by Ludwig Boltzmann, (M.P.P., the editor of the book) was told a few months the theories of spin glasses and Hopfield's models after its publication, during a visit at MIT, that it was soon began to converge, jointly establishing many `obvious' (sic, a qualification whose importance will basic analogies between idealized magnetic lattices be clear in a moment) that Chomsky and Fodor had and idealized neural nets. The formal equivalence won the debate. A few weeks later, in Geneva, he was between spin glasses and neuronal networks, no told that it was `obvious' (again, sic) that Piaget, matter how elegant and intellectually satisfactory at Inhelder, CeÂlleÂrier and Papert had won the debate. an abstract level, engendered in some neurobiolo- The self-imposed neutrality of MPP while editing the gists doubts concerning the real applicability of proceedings was thus powerfully, albeit indirectly, these models to real brain circuits. acknowledged. Many years later, this neutrality was 37. At the time (and until the mid-1990s) the nature of abandoned in an afterthoughts piece in Cognition these parameters was conceived as quintessentially (1994), decidedly in favor of Chomsky and Fodor. syntactic. Under the guidance of specific hypotheses But opposite conclusions and afterthoughts have about the parameters, researchers applied real acqui- been reached in other quarters. Those who have sition data (and vast precompiled corpora) to their sided with Piaget typically accuse Chomsky and theoretical predictions of the subtle cascades of mani- Fodor of having been quite ungracious in rejecting fest consequences on linguistic expressions that the Piaget's many overtures and concessions to their different possible parametric values were expected positions, and of having countered his simple, unten- to produce. This kind of linguistico-developmental dentious and unassailable theses with a flurry of `parametric' literature started to grow steadily (and possibly (just possibly) relevant paradoxes and con- still continues to the present date), but the early undra, for which (so the story goes) they had no results were not always neat, and many intense dis- solutions to offer. Present and future readers will cussions ensued. In 1995, the `minimalist program' have to decide for themselves which conclusion is (Chomsky, 1995) shifted the localization of the par- more correct. ameters from the syntax proper to the morpho-lexical 35. The theories of conceptual development, categoriza- component of language. The problematic reinterpret- tion (Rosch, 1973, 1978, 1996; Smith et al., 1988; Smith ation of older data and a new flurry of experiments and Medin, 1981) and psychological similarity (She- carried out under this different model have blurred pard, 1962, 1964, 1994; Tversky, 1977; Tversky and some of the simple and elegant contours of the initial Gati, 1978; see the special issue of Brain and Behavioral experiments. This has stimulated some, but also dis- Sciences on the work of Roger Shepard for a recent appointed other, researchers, who presently feel the synthesis) combined with the theories of, and experi- need for a reappraisal of the very ideas behind the PP ments on, lexical acquisition in the child, and with model. Whatever the ultimate destiny of the PP ideal- lexical semantics, opened the way to an integrated ization, it has given a very productive impulse to the theory of conceptual and linguistic development at development of detailed non-inductivist models of the interfaces between phonology and syntax, syntax learning. Novel kinds of fixation mechanisms and and the lexicon, syntax and semantics (Bloom, 2000; equally novel kinds of interactions between the lin- Gopnik and Meltzoff, 1998; Jackendoff, 1983, 1990; guistic external inputs and these mechanisms have Landau and Gleitman, 1985; Levin and Pinker, 1991; been introduced. One quick and relevant measure is with a dissenting view by Fodor, 1994, 1998; also see a steady increase in the number of scientific commu- Phonology; Phonological Processes; Syntax; Syntax, nications inspired by this approach presented each Aquisition of; Syntax and Semantics: Formal Ap- year at the Boston University Conference on Lan- proaches). guage Development. The Conferences started in 36. It is noteworthy that, since the mid-1980s, physicists 1976. Their proceedings from 1994 to the present in France (Marc MeÂzard and GeÂrard Toulouse), in have been published by Cascadilla Press (Somerville, Israel (Daniel Amit and Gabriele Veneziano) and Massachusetts). in Italy and Argentina (Giorgio Parisi and Miguel An entire family of new mechanisms was pro- Virasoro) have developed germane mathematical posed to account for non-inductive learning and the models for the so-called spin glasses (amorphous fixation and stabilization of rules and concepts in the magnetic lattices in which each node is occupied child's developing mind±brain: the subset principle xxxii What is Cognitive Science?

(Berwick, 1985), greediness, conservativity, locking A series of large multidisciplinary conferences on capacity, triggers, and default values (Wexler and consciousness has been held at two-year intervals in Culicover, 1980; Wexler and Borer, 1986; Wexler Tucson, Arizona since 1994, each time attracting and Manzini, 1986; Gibson and Wexler, 1994). A more than a thousand participants ranging from close constructive dialogue was established between poets to physicists to physiologists. formal theories of `learnability' (Gold, 1967; Osher- 42. The 1990 article in Brain and Behavioral Sciences by son et al., 1986; Pinker, 1979) and empirical data on Pinker and Bloom has been received with warm, language acquisition in children in a variety of lan- albeit sometimes cautious (Deacon, 1997), assent by guages (for an early synthesis, see Wanner and Gleit- many who have only general sympathy for genera- man, 1982; for a recent one see Guasti, 2002). tive grammar, or accept only parts of it, or remain Arguably, the most remarkable property of these prudently agnostic about the enterprise as a whole. hypotheses, besides their non-inductivism, is that The reconciliation between generative grammar and they constitute a cumulative attempt by many of a neo-Darwinian adaptationist account of the evolu- these researchers to build a detailed selectional (as tion of language, so eagerly explored by Pinker and opposed to instructional) frame for language acquisi- Bloom (for later developments and refinements see tion across languages. Selective models of the growth their subsequent books: Bloom, 2000; Pinker, 1994, and development of the nervous system can be 1997), reassures those who are reluctant to follow traced back to the paradigm of stabilization of de- Chomsky all the way in his defense of radical discon- veloping synapses through selective activation, as tinuity and the punctate appearance of the language initially established by Hubel and Wiesel's experi- faculty exclusively in humans. It assuages their fear ments on the development of the visual system in that Chomsky's theses may involve an appeal to an the selectively deprived kitten. Pasko Rakic and Pa- evolutionary miracle, an exceedingly improbable tricia Goldman-Rakic (Goldman-Rakic, 1985; Rakic `hopeful monster'. The radical adversaries of genera- et al., 1986) and Purves and Lichtman, among others, tive grammar, predictably, have found no reason to later extended and refined a model based on the be interested in this reconciliation, noting that the overproduction of synapses in the developing cortex, generative camp harbors embarrassing internal dis- followed by a massive trimming of the inactive con- agreements. Generative grammarians have, in the nections. The wider implications of such selective main, remained unabatedly critical of all extant models for neurobiology, cognitive science and adaptationist accounts of language evolution (see beyond were soon highlighted by J.-P. Changeux note 25), including the one offered by Pinker and and G. M. Edelman (Changeux et al., 1984; Edelman Bloom (see the peer commentaries accompanying and Mountcastle, 1978; Edelman and Reeke, 1982; the Brain and Behavioral Sciences article), and some for a recent reconstruction and wide philosophical have brought into evidence the convergent critiques consequences of selective theories, see Changeux, of adaptationist explanations independently and au- 2002). thoritatively developed by Stephen J. Gould and It is to be expected that at least some of these new Richard C. Lewontin in evolutionary theory proper ideas will find a suitable place in future theories of (one of the main targets of Pinker and Bloom was a learning and acquisition, possibly even beyond the paper by Piattelli-Palmarini (1989) in which this domain of language. Over and beyond the interest of convergence between antecedently separate contri- the PP model per se, this is a case study of a deep butions was made explicit). At an international meet- connection between the neurosciences and the study ing in Venice (Italy), with Gould in the audience, of the mind, one of many that have shaped modern Paul Bloom confessed with humor that he and Pinker cognitive science. would have felt uncomfortable challenging Chomsky 38. Many have contributed to this set of exciting devel- on language and Gould on evolution, but felt reason- opments, but the St. Louis group of Marcus Raichle, ably comfortable challenging Chomsky on evolution Peter Fox, Steve Peterson and Michael Posner, instru- and Gould on language. mental in getting things started, deserve special men- tion (Posner et al., 1988). REFERENCES 39. This combination is currently being explored most productively by Nikos Logothetis and his group at Adolphs R, Tranel D, Bechara A, Damasio H and the Max Planck Instiutute in TuÈ bingen (Saleem et al., Damasio AR (1996) Neuropsychological approaches 2002). to reasoning and decision-making. In: Damasio AR 40. For instance, among many more, the following art- (ed.) Neurobiology of Decision-making. New York: icles: The New York Times, March 30, 1997; The New Springer. York Times Magazine, February 11, 2001; The Washing- Albertazzi L (2001) The Dawn of Cognitive Science: Early ton Post, January 27, 2002. European Contributors. Dordrecht: Kluwer. 41. This resurrection was initiated perhaps by research- Ashby WR (1960) Design for a Brain: The Origins of ers in cognitive psychology and neuropsychology Adaptive Behavior. New York: Wiley. (e.g. Moscovitch, 1995), but it has now been em- Baars BJ (1986) The Cognitive Revolution in Psychology. braced by a much wider interdisciplinary network. New York: The Guilford Press. What is Cognitive Science? xxxiii

Bar-Hillel Y (1953a) On recursive definitions in empirical Bliss TVP and Lomo T (1973) Long-lasting potentiation of sciences. In: Proceedings of the 11th Congress of synaptic transmission in the dentate area of the Philosophy, Brussels V, pp. 160±165. anaesthetised rabbit following stimulation of the Bar-Hillel Y (1953b) A quasi-arithmetical notation for perforant path. Journal of Physiology 232: 331±356. syntactic description. Language 29: 47±58. Block N (1978) Troubles with functionalism. In: Savage Bar-Hillel Y (1954) Logical syntax and semantics. CW (ed.) Perception and Cognition: Issues in the Language 30: 230±237. Foundations of Psychology: Minnesota Studies in the Bartlett FC (1932) Remembering: a Study in Experimental Philosophy of Science, vol. 9. Minneapolis: University and Social Psychology. Cambridge, UK: Cambridge of Minnesota Press. University Press. Block N (ed.) (1980) Readings in the Philosophy of Barton EG, Berwick RC and Ristad ES (1987) Psychology, vol. 1. Cambridge, MA: Harvard University Computational Complexity and Natural Language. Press. Cambridge, MA: Bradford Books/MIT Press. Block N (ed.) (1981) Readings in the Philosophy of Bates E and Dick F(2000) Beyond phrenology: brain and Psychology, vol. 2. Cambridge, MA: Harvard language in the next millennium. Brain and Language University Press. 71: 18±21. Bloom P (2000) How Children Learn the Meanings of Words. Bates E and Elman J (1996) Learning rediscovered. Science Cambridge, MA: Bradford Books/MIT Press. 274: 1849±1850. Boller F(1998) History of the International Bates E, Elman J, Johnson M et al. (1999) Innateness and Neuropsychological Symposium: a reflection of the emergentism. In: Bechtel W and Graham G (eds) A evolution of a discipline. Neuropsychologia 37: 17±26. Companion to Cognitive Science. Malden, MA and Bower B (2002) A maverick reclaimed: some Oxford: Blackwell. psychologists say it's time that Egon Brunswik got his Beach FA, Hebb DO, Morgan CT and Nissen HW (eds) due. Science News 161: 155±156. (1960) The Neuropsychology of Lashley: Selected Papers of Braun M (1992) Picturing Time: The Work of Etienne-Jules K. S. Lashley. New York: McGraw-Hill. Marey. Chicago, IL: University of Chicago Press. Bechara A, Damasio AR and Damasio H (1994) Brazier MAB (ed.) (1958) The Central Nervous System and Insensitivity to future consequences following damage Behavior: Transactions of the First Conference. New York: to human prefrontal cortex. Cognition 50: 7±15. Josiah Macy Jr. Foundation. Bechara A, Damasio H, Damasio AR and Lee GP (1999) Brazier MAB (ed.) (1959) The Central Nervous System and Different contribution of the human amygdala and Behavior: Transactions of the Second Conference. ventromedial prefrontal cortex to decision-making. New York: Josiah Macy Jr. Foundation. The Journal of Neuroscience 19(13): 5473±5481. Brazier MAB (ed.) (1960) The Central Nervous System and Bechtel W, Abrahamsen A and Graham G (1999) The life Behavior: Transactions of the Third Conference. New York: of cognitive science. In: Bechtel W and Graham G (eds) Josiah Macy Jr. Foundation. ACompanion to Cognitive Science , pp. 1±104. Malden, Breiter HC, Aharon I, Kahneman D, Dale A and Shizgal P MA and Oxford: Blackwell. (2001) Functional imaging of neural responses to Bechtel W and Graham G (eds) (1999) ACompanion to expectancy and experience of monetary gains and Cognitive Science. Malden, MA and Oxford: Blackwell. losses. Neuron 30: 619±639. Bellugi U, Lichtenberger L, Mills D, Galaburda AM and Broadbent D (1958) Perception and Communication. Korenberg JR (1999) Bridging cognition, brain and New York: Pergamon Press. molecular genetics: evidence from Williams syndrome. Broca P (1878) Anatomie compareÂe des circomvolutions Trends in Neurosciences 22(5): 197±207. ceÂreÂbrales. Le grand lobe limbique et la scissure Bellugi U and St. George M (2000) Linking cognitive limbique dans la seÂrie des mammifeÁres. Revue neuroscience and molecular genetics: new anthropologique 1: 385±498. perspectives from Williams syndrome. Journal of Brown R (1958) How shall a thing be called? Psychological Cognitive Neuroscience 12(1): 1±107. Review 65: 14±21. Bellugi U, St. George M and Galaburda AM (eds) (2001) Brown R (1973) AFirst Language: The Early Stages . Journey from Cognition to Brain to Gene: Perspectives from Cambridge, MA: Harvard University Press. Williams Syndrome. Cambridge, MA: Bradford Books/ Bruner JR, Goodnow JJ and Austin GA (1956) AStudy MIT Press. of Thinking. New York: Wiley. Bellugi U, Wang P and Jernigan TL (1994) Williams Brush SG (1967) History of the Lenz±Ising model. Review syndrome: An unusual neuropsychological profile. In: of Modern Physics 39: 883±893. Broman S and Grafman J (eds) Atypical Cognitive Carey S (1985) Conceptual Change in Childhood. Deficits in Developmental Disorders: Implication for Brain Cambridge, MA: Bradford Books/MIT Press. Function, pp. 23±56. Mahwah, NJ: Erlbaum. Changeux J-P (2002) L'homme de veÂrite. Paris: Odile Jacob. Berwick RC (1985) The Acquisition of Syntactic Knowledge. Changeux J-P (ed.) L'unite de l'homme: invariants Cambridge, MA: MIT Press. biologiques et universaux culturels. Paris: Seuil. Blakemore C and Cooper RM (1970) Development of the Changeux J-P, Heidmann T and Patte P (1984) Learning brain depends upon the visual environment. Nature by selection. In: Marler P and Terrace HS (eds) The 228: 477±478. Biology of Learning. New York: Springer Verlag. xxxiv What is Cognitive Science?

Cherry C (ed.) (1961) Information Theory: Fourth London Delafresnaye JF(ed.) (1954) Brain Mechanims and Symposium. London, UK: Butterworths. Consciousness: ASymposium Organized by the Council for Chomsky N (1955) The Logical Structure of Linguistic International Organizations of Medical Sciences Theory. Cambridge, MA: MIT Press. (C.I.O.M.S.). Springfield, IL: Charles C Thomas. Chomsky N (1956) Three models for the description of Delafresnaye JF(ed.) (1961) Brain Mechanims and language. In: Luce RD, Bush RR and Galanter E (eds) Learning: ASymposium Organized by the Council for Readings in Mathematical Psychology, vol. 2, pp. 105±124. International Organizations of Medical Sciences New York: John Wiley and Sons. (C.I.O.M.S.). Oxford, UK: Blackwell. Chomsky N (1957) Syntactic Structures. The Hague: Dennett DC (1987) The Intentional Stance. Cambridge, Mouton. MA: Bradford Books/MIT Press. Chomsky N (1959/1980) A review of B. F. Skinner's Dunbar R (1999) Grooming, Gossip and the Evolution of `Verbal Behavior'. Language 35(1): 26±58. Language. Cambridge, MA: Harvard University Chomsky N (1965) Aspects of the Theory of Syntax. Press. Cambridge, MA: MIT Press. Dupuy J-P (2000) The Mechanization of the Mind: On the Chomsky N (1975) The Logical Structure of Linguistic Origins of Cognitive Science. Princeton, NJ: Princeton Theory. Chicago, MA: University of Chicago Press. University Press. Chomsky N (1980) Rules and Representation. Columbia Edelman GM (1987) Neural Darwinism: The Theory of University Press. Neuronal Group Selection. New York: Basic Books. Chomsky N (1981) Lectures on Government and Binding Edelman GM and Mountcastle VB (1978) The Mindful (The Pisa Lectures). Dordrecht: Foris. Brain: Cortical Organization and the Group-selective Chomsky N (1986) Knowledge of Language: Its Nature, Theory of Higher Brain Function. Cambridge, MA: Origin, and Use. New York: Praeger Scientific. Bradford Books/MIT Press. Chomsky N (1995) Language and nature. Mind 104: Edelman GM and Reeke GN (1982) Selective networks 1±61. capable of representative transformations, limited Chomsky N (1995) The Minimalist Program. Cambridge, generalizations, and associative memory. Proceedings MA: MIT Press. of the National Academy of Sciences of the USA 79: Chomsky N (1998) On Language. New York: New Press. 2091±2095. Chomsky N (ed.) (1988) Language and Interpretation: Elman JL (1989) Representation and Structure in Philosophical Reflections and Empirical Inquiry. Connectionist Models, Center for Research in Language Chomsky N and Miller GA (1963) Introduction to the Technical Report 8903. La Jolla, CA: University of formal analysis of natural languages. In: Luce RD, California. Bush RR and Galanter E (eds) Handbook of Mathematical Elman JL, Bates EA, Johnson MH et al. (1996) Psychology, vol. 2, pp. 269±321. New York: John Wiley Rethinking Innateness: AConnectionist Perspective on and Sons. Development. Cambridge, MA: Bradford Books/MIT Church A (1936) An unsolvable problem of elementary Press. number theory. The American Journal of Mathematics Emmorey K and Lane H (eds) (2000) The Signs of Language 58: 345±363. Revisited: An Anthology in Honor of Ursula Bellugi and Cohen NJ and Squire LR (1980) Preserved learning and Edward Klima. Mahwah, NJ: Lawrence Erlbaum. retention of pattern-analyzing skill in amnesia: Figueredo AJ (1992) Preparedness and plasticity: a dissociation of knowing how and knowing that. Science stochastic optimality theory. Human Behavior and 210: 207±210. Evolution Society: Fourth Annual Meeting. Albuquerque, Cowie F(1999) What's Within? Nativism Reconsidered. NM. New York and Oxford, UK: Oxford University Press. Fodor J and Garrett M (1966) Some reflections on Cowie F(2000) Whistling `Dixie': response to Fodor.In: competence and performance (with comments by Nani M and Marraffa M (eds) E-Symposium on Fiona N Stuart Sutherland and L Jonathan Cohen). In: Lyons Cowie's `What's Within?: Nativism Reconsidered' (SIFA J and Wales RJ (eds) Psycholinguistics Papers E-Symposia). [http://host.uniroma3.it/progetti/kant/ (Proceedings of the Edinburgh Conference 1966), field/booksymp.html.] pp. 135±179. Edinburgh, UK and Chicago, IL: Craik K (1943) The Nature of Explanation. Cambridge, UK: Edinburgh University Press/Aldine Publishing Cambridge University Press. Company. Damasio AR (1994) Descartes' Error. Grossett Putnam. Fodor JA (1975) The Language of Thought. New York: Damasio AR, Damasio H and Christen Y (1996) Thomas Y Crowell. Neurobiology of Decision-making. Springer. Fodor JA (1981) Representations. Harvester Press. Davis M (ed.) (1965) The Undecidable: Basic Papers on Fodor JA (1983) The Modularity of Mind: An Essay on Undecidable Propositions, Unsolvable Problems and Faculty Psychology. Cambridge, MA: Bradford Computable Functions. Hewlett, NY: Raven Press. Books/MIT Press. De Reuck AVS and O'Connor M (eds) (1964) Disorders of Fodor JA (1990) ATheory of Content and Other Essays . Language. Boston, MA: Little, Brown and Co. Cambridge, MA: Bradford Book/MIT Press. Deacon TW (1997) The Symbolic Species. London: WW Fodor JA (1994) Concepts: a potboiler. Cognition 50: Norton and Company Inc. 95±113. What is Cognitive Science? xxxv

Fodor JA (1998) Concepts: Where Cognitive Science Went Goldman-Rakic PS (1987) Circuitry of primate prefrontal Wrong. New York and Oxford, UK: Oxford University cortex and the regulation of behavior by representation Press. memory. In: Phern F(ed.) Handbook of Physiology, vol. 5. Fodor JA, Bever TG and Garrett MF (1974) Gopnik A and Meltzoff AN (eds) (1998) Words, Thoughts, The Psychology of Language: An Introduction to and Theories. Cambridge, MA: MIT Press. Psycholinguistics and Generative Grammar. New York: Gopnik M (1990) Feature-blind grammar and dysphasia McGraw-Hill. 344: 715. Fodor JA and Lepore E (1992) Holism. New York and Gopnik M (1994) Impairments of tense in a familial Oxford, UK: Blackwell. language disorder. Journal of Neurolinguistics 8(2): Fouts RS and Fouts DH (1989) Loulis in conversation 109±133. with cross-fostered chimpanzees. In: Gardner RA, Gould E and Gross CG (2002) Neurogenesis in adult Gardner BT and Van Cantfort TE (ed.) Teaching mammals: some progress and problems. The Journal of Sign Language to Chimpanzee. Albany, NY: CUNY Neuroscience 22: 619±623. Press. Gross CG, Rocha-Miranda CE and Bender D (1972) Frege G (1879) Begriffsschrift, eine der arithmetischen Visual properties of neurons in inferotemporal nachgebildete formelsprache des reinen denken. Halle. cortex of the macaque. Journal of Neurophysiology 35: Gaffan D (1974) Recognition impaired and association 96±111. intact in the memory of monkeys after transection of Grossberg S (1976) Adaptive pattern classification and the fornix. Journal of Comparative and Physiological universal recoding: 1. parallel development and Psychology 86: 1100±1109. coding of neural feature detectors. Biological Cybernetics Gardner BT and Gardner RA (1971) Behavior of Non- 23: 121±134. Human Primates, vol. 4. New York: Academic Press. Guasti MT (2002) Language Acquisition: The Growth of Gardner H (1985) The Mind's New Science: AHistory of the Grammar. Cambridge, MA: MIT Press. Cognitive Revolution. New York: Basic Books. Haas RB (1976) Muybridge: Man in Motion. Berkeley, CA: Gardner RA, Gardner BT and Van Cantfort TE (ed.) University of California Press. (1989) Teaching Sign Language to Chimpanzee. Albany, Hammond KR and Stewart TR (eds) (2001) The Essential NY: CUNY Press. Brunswik: Beginnings, Explications, Applications.New Gauthier I, Skudlarsky P, Gore JC and Anderson AW York and Oxford, UK: Oxford University Press. (2000a) Expertise for cars and birds recruits brain areas Hanson SJ (1999) Connectionist neuroscience: involved in face recognition. Nature Neuroscience 3: representational and learning issues for neuroscience. 191±197. In: Lepore E and Pylyshyn (eds) What is Cognitive Gauthier I, Tarr MJ, Anderson AW, Skudlarsky P and Science?, pp. 401±427. Malden, MA: Blackwell. Gore JC (1999) Activation of the middle fusiform 'face Harris ZS (1951) Methods of Structural Linguistics. area' increases with expertise in recognizing novel Chicago, IL: University of Chicago Press. objects. Nature Neuroscience 2: 568±573. Harris ZS (1952a) Discourse analysis (Part 1). Language Gauthier I, Tarr MJ, Moylan J et al. (2000b) The fusiform 28: 18±23. `face area' is part of a network that processes faces at Harris ZS (1952b) Discourse analysis (Part 2): A sample the individual level. Journal of Cognitive Neuroscience 12: text. Language 28: 474±494. 495±504. Harris ZS (1957) Co-occurrence and transformation in Gazzaniga MS (1970) The Bisected Brain. New York: linguistic structure. Language 33: 293±340. Appleton-Century-Crofts. Harris ZS (1986/1951) Structural Linguistics. Chicago, IL: Gazzaniga MS (1984) Handbook of Cognitive Neuroscience. University of Chicago Press. New York: Plenum Press. Haugeland J (1997) What is mind design. In: Gazzaniga MS (ed.) (2000) The New Cognitive Haugeland J (ed.) Mind Design II: Philosophy, Neurosciences, 2nd edn. Cambridge, MA: Bradford Psychology, Artificial Intelligence, pp. 1±28. Cambridge, Books/MIT Press. MA: Bradford Books/MIT press. [revised and Gibson E and Wexler K (1994) Triggers. Linguistic Inquiry enlarged edition] 25: 407±454. Haugeland J (ed.) (1981) Mind Design: Philosophy, Gigerenzer G, Hoffrage U and KleinboÈlting H (1991) Psychology, Artificial Intelligence. Cambridge, MA: Probabilistic mental models: a Brunswikian theory of Bradford Books/MIT press. confidence. Psychological Review 98(4): 506±528. Hauser MD, Chomsky N and Fitch WT (in press) The Gold EM (1967) Language identification in the limit. faculty of language: what it is, who has it, and how did Information and Control 10: 447±474. it evolve? Science. Goldman AI and Gallese V (1998) Mirror neurons and the Hebb DO (1949) The Organization of Behavior: simulation theory of mind-reading. Trends in Cognitive ANeuropsychological Theory . New York: Science 2(12): 493±501. Wiley-Interscience. Goldman-Rakic P (1985) Toward a neurobiology of Heims SJ (1991) The Cybernetics Group. Cambridge, MA: cognitive development. In: Mehler RFJ (ed.) Neonate MIT Press. Cognition: Beyond the Blooming Buzzing Confusion. Higginbotham JT (1982) Noam Chomsky's linguistic Hillsdale, NJ: Lawrence Erlbaum Associates. theory. Social Research 49(1): 143±157. xxxvi What is Cognitive Science?

Hirsh R (1974) The hippocampus and contextual retrieval Kanwisher N and Moscovitch M (2000) The cognitive of information from memory: a theory. Behavioural neuroscience of face processing: an introduction. Biology 12: 421±444. Cognitive Neuropsychology 1/2/3: 1±13. Hochberg J (1956) Perception: toward the recovery of a Karmiloff K and Karmiloff-Smith A (2001) Pathways to definition. Psychological Review 63: 400±405. Language: From Fetus to Adolescent. Cambridge, MA: Hochberg J (1978) Perception. Englewood Cliffs, NJ: Harvard University Press. Prentice-Hall. Karmiloff-Smith A (1992) Beyond Modularity: A Hopfield JJ (1982) Neural networks and physical systems Developmental Perspective on Cognitive Science. with emergent collective computational abilities. Cambridge, MA: MIT Press. Proceedings of the National Academy of Sciences of the Karmiloff-Smith A (1993) Beyond Modularity: a U.S.A 81: 2554±2558. developmental perspective on cognitive science. Hubel DH and Wiesel TN (1962a) Receptive fields, European Journal of Disorders of Communication 28: binocular interaction and functional architecture in the 95±105. cat's visual cortex. Journal of Physiology 160: 106±154. Karmiloff-Smith A (1994) PreÂcis of beyond modularity: Hubel DH and Wiesel TN (1962b) Receptive fields of a developmental perspective on cognitive science. single neurons in the cat's striate cortex. Journal of Behavioral and Brain Sciences 17: 693±745. Physiology 148: 574±591. Karmiloff-Smith A (1998) Development itself is the key to Hull CL (1943) Principles of Behavior. New York: understanding developmental disorders. Trends in Appleton-Century-Crofts. Cognitive Sciences 2(10): 389±398. Jackendoff R (1983) Semantics and Cognition. Cambridge, Kegl J and McWorther J (in press) Perspectives on an MA: MIT Press. emerging language. Proceedings of the 28th Annual Jackendoff R (1990) Semantic Structures. Cambridge, MA: Stanford Child Language Research Forum. New York and MIT Press. Cambridge UK: Cambridge University Press. Jackendoff RP, Bloom P and Wynn K (eds) (1999) Kegl J, Neidle C, MacLaughlin D, Bahan B and Lee RG Language, Logic, and Concepts: Essays in Honor of John (2000) The Syntax of American Sign Language: Functional MacNamara. Cambridge, MA: MIT Press. Categories and Hierarchical Structure. Cambridge, MA: James WJ (1890) The Principles of Psychology. New York: MIT Press. Henry Holt. Keil FC (1979) Semantic and Conceptual Development: An Jasper HH and Smirmov GD (eds) (1960) The Moscow Ontological Perspective. Cambridge, MA: Harvard colloquium on the electroencephalography of higher University Press. nervous activity. The International Journal of Kinsbourne M and Wood F(1975) Short-term memory Electroencephalography and Clinical Neurophysiology and pathological forgetting. In: Deutsch D and Deutsch 13(suppl.). AJ (eds) Short-term Memory. New York: Academic Jeffress LA (ed.) (1951) Cerebral Mechanisms in Behavior: Press. The Hixon Symposium. New York: John Wiley. Kleene SC (1936) General recursive functions of natural Jenkins L (2000) Biolinguistics: Exploring the Biology of numbers. Mathematische Annalen 112: 727±742. Language. Cambridge: Cambridge University Press. Klima ES and Bellugi U (1979) The Signs of Language. Johansson G (1973) Visual perception of biological Cambridge, MA: Harvard University Press. motion and a model for its analysis. Perception and Kosslyn SM (1980) Image and Mind. Cambridge, MA: Psychophysics 14(2): 201±211. Harvard University Press. Kahneman D, Slovic P and Tversky A (eds) (1982) Kosslyn SM and Andersen RA (eds) (1992) Frontiers in Judgment Under Uncertainty: Heuristics and Biases. Cognitive Neuroscience. Cambridge, MA: Bradford Cambridge, UK and New York: Cambridge University Books/MIT Press. Press. Kubie L (1930) A theoretical application to some Kahneman D and Tversky A (1972) Subjective neurological problems of the propensities of excitation probability: a judgment of representativeness. waves which move in closed circuits. Brain 53: 166±177. Cognitive Psychology 3: 43±454. Landau B and Gleitman LR (1985) Language and Kahneman D and Tversky A (1973) On the psychology of Experience: Evidence from the Blind Child. Cambridge, prediction. Psychological Review 80: 237±251. MA: Harvard University Press. Kahneman D and Tversky A (1979) Prospect theory: Larson R and Segal G (1995) Knowledge of Meaning: An an analysis of decision under risk. Econometrica 47: Introduction to Semantic Theory. Cambridge, MA: MIT 263±291. Press. Kamin LJ (1969) Predictability, surprise, attention and Lashley KS (1929) Brain Mechanisms and Intelligence. conditioning. In: Campbell BA and Church RM (eds) Chicago, IL: University of Chicago Press. Punishment and Aversive Behavior, pp. 279±296. Lashley KS (1951) The problem of serial order in New York: Appleton-Century-Crofts. behavior. In: Jeffress LA (ed.) Cerebral Mechanisms in Kandel ER (1980) Cellular insights into the multivariant Behavior, pp. 112±136. New York: John Wiley. nature of arousal. In: McFadden D (ed.) Neural Lettvin JY, Maturana HR, McCulloch WS and Pitts WH Mechanisms in Behavior, pp. 260±291. New York: (1959) What the frog's eye tells the frog's brain. Springer-Verlag. Proceedings of the IRE 47: 1940±1951. What is Cognitive Science? xxxvii

Levin B and Pinker S (eds) (1991) Lexical and Conceptual Mountcastle VB (1957) Modality and topographic Semantics. New York and Oxford: Blackwell. properties of single neurons of cat's somatic sensory Lewis HR and Papadimitriou CH (1981) Elements of the cortex. Journal of Neurophysiology 20: 408±434. Theory of Computation. Prentice-Hall. Nadel L and O'Keefe J (1974) The hippocampus in pieces Lieberman P (2000) Human Language and Our Reptilian and patches: an essay on modes of explanation in Brain: The Subcortical Bases of Speech, Syntax, and physiological psychology. In: Bellairs R and Gray EG Thought. Cambridge, MA: Harvard University Press. (eds) Essays on the Nervous System. AFestschrift for J.Z. Lightfoot D (1982) The Language Lottery: Toward a Biology Young. Oxford, UK: Clarendon Press. of Grammars. MIT Press. Neisser U (1967) Cognitive Psychology. Englewood Cliffs: Lorente de No R (1938) Analysis of the activity of the Prentice-Hall. chains of internuncial neurons. Journal of Nespor M (2001) About parameters, prominence, and Neurophysiology 1: 207±244. bootstrapping. In: Dupoux E (ed.) Language, Brain, and Mackintosh NJ (1975) A theory of attention: variations in Cognitive Development, pp. 127±142. Cambridge, MA: the associability of stimuli with reinforcements. Bradford Books/MIT Press. Psychological Review 82: 276±298. Newell A, Shaw JC and Simon HA (1958) Elements of a Markman EM (1989) Categorization and Naming in theory of human problem-solving. Psychological Review Children: Problems of Induction. Cambridge, MA: 65: 151±166. Bradford Books/MIT Press. Newell A and Simon HA (1972) Human Problem Solving. Marr D (1982) Vision: AComputational Investigation into Englewood Cliffs, NJ: Prentice-Hall. the Human Representation and Processing of Visual Newport EL, Bavelier D and Neville HJ (2001) Critical Information. San Francisco, CA: WH Freeman. thinking about critical periods: perspectives on a McCabe KA, Houser D, Ryan L, Smith VL and Trouard T critical period for language acquisition. In: Dupoux E (2001) A functional imaging study of cooperation in (ed.) Languge, Brain and Cognitive Development: Essays in two-person reciprocal exchange. Proceedings of the Honour of Jacques Mehler, pp. 481±502. Cambridge, MA: National Academy of Sciences of the USA 98: 11832±11835. Bradford Books/MIT Press. McCulloch WS (1964) What's in the brain that ink may Nowak MA, Komarova NL and Niyogi P (2002) character? The International Congress for Logic, Computational and evolutionary aspects of language. Methodology and Philosophy of Science (Jerusalem, Israel, Nature 417: 611±617. August 26). O'Keefe J and Dostrovsky J (1971) The hippocampus McCulloch WS (1988) Embodiments of Mind. Cambridge, as a spatial map: preliminary evidence from unit MA: MIT Press. activity in the freely-moving rat. Brain Research 34: McCulloch WS and Pitts WH (1943) A logical calculus of 171±175. the ideas immanent in nervous activity. Bulletin of O'Keefe J and Nadel L (1978) The Hippocampus as a Mathematical Biophysics 5: 115±133. Cognitive Map. Oxford, UK: Clarendon Press. MeÂzard M, Parisi G and Virasoro MA (1987) Spin Glass O'Keefe J and Nadel L (1979) The hippocampus Theory and Beyond. Singapore and Teaneck, NJ: World as a cognitive gap. Behavioral and Brain Sciences 2(4). Scientific. Olds J and Milner PM (1954) Positive reinforcement Miller GA and Chomsky N (1963) Finitary models of produced by electrical stimulation of the septal area language users. In: Luce RD, Bush RR and Galanter E and other regions of the rat brain. Journal of Comparative (eds) Handbook of Mathematical Psychology, vol. 2, and Physiological Psychology 47: 419±427. pp. 419±491. New York: John Wiley and Sons. Osgood CE and Miron MS (eds) (1963) Approaches to the Miller GA, Galanter E and Pribram KH (1960) Plans and Study of Aphasia. A Report of an Interdisciplinary the Structure of Behavior. New York: Holt, Rinehart and Conference on Aphasia. Urbana, IL: University of Illinois Winston. Press. Milner B (1965) Memory disturbance after bilateral Osherson DN (ed.) (1990/1995) An Invitation to Cognitive hippocampal lesions. In: Milner PM and Glickman SE Science. Cambridge, MA: Bradford Books/MIT Press. (eds) Cognitive Processes and the Brain, pp. 97±111. [1st edn (1990) in 3 volumes. 2nd edn (1995) in Patterson, NJ: Van Nostrand. 4 volumes.] Minsky M (1961) Steps toward Artificial Intelligence. Osherson DN, Stob M and Weinstein S (1986) Systems that Proceedings of IRE 49: 8±30. Learn. Cambridge, MA: MIT Press. Minsky ML and Papert SA (1990) Perceptrons.MIT Penfield W and Rasmussen T (1949) Vocalization and Press. arrest of speech. Archives of Neurology and Psychiatry Moruzzi G and Magoun HW (1949) Brain stem reticular 61: 21±27. formation and activation of the EEG. Peterson MA (1999) Top-down influences on distinctly Electroencephalography and Clinical Neurophysiology 1: perceptual processes. Behavioral and Brain Sciences 455±473. 22: 389±390. Moscovitch M (1995) Recovered consciousness: a Petitto LA (1987) On the autonomy of language and hypothesis concerning modularity and episodic gesture: Evidence from the acquisition of personal memory. Journal of Clinical and Experimental pronouns in American Sign Language. Cognition 27(1): Neuropsychology 17: 276±290. 1±52. xxxviii What is Cognitive Science?

Petitto LA and Marentette P (1991) Babbling in the Pribram KH and Gill MM (1976) Freud's 'Project' Re- manual mode: Evidence for the ontogeny of language. assessed : Preface to Contemporary Cognitive Theory and Science 251: 1483±1496. Neuropsychology. New York: Basic Books. Piaget J (1930) The Child's Conception of the World.New Putnam H (1960) Minds and Machines. In: Collected York: Harcourt Brace. Papers. Vol. 2: Mind, Language and Reality, pp. 362±384. Piaget J (1954) The Construction of Reality in the Child. New York and Cambridge, UK: Cambridge University Ballantine Books. Press. Piattelli-Palmarini M (1989) Evolution, selection and Putnam H (1973) Philosophy and our mental life. In: cognition: from `learning' to parameter setting in Mind, Language and Reality. Philosophical Papers: Vol 2, biology and in the study of language. Cognition 31: pp. 291±303. Cambridge: Cambridge University 1±44. Press. Piattelli-Palmarini M (1994) Ever since language and Putnam H (1987) The Many Faces of Realism. La Salle, IL: learning: after thoughts on the Piaget±Chomsky Open Court. debate. Cognition 50: 315±346. Putnam H (1988) Representation and Reality. Cambridge, Piattelli-Palmarini M (2001) Portrait of a `classical' MA: MIT Press/Bradford Books. cognitive scientist: what I have learned from Jacques Quine WVO (1956) Quantifiers and propositional Mehler. In: Dupoux E (ed.) Language, Brain, and attitudes. Journal of Philosophy 53(5): 177±187. Cognitive Development, pp. 3±21. Cambridge: Bradford Rakic P, Bourgeois J-P, Eckenhoff MF, Zecevic N and Books/MIT Press. Goldman-Rakic P (1986) Concurrent overproduction of Piattelli-Palmarini M (ed.) (1980) Language and synapses in diverse regions of the primate cerebral Learning: The Debate between Jean Piaget and Noam cortex. Science 232: 232±234. Chomsky. Cambridge, MA: Harvard University Ramus F(2002) Language discrimination by newborns: Press. teasing apart phonotactic, rhythmic, and intonational Pinker S (1979) Formal models of language learning. cues. Annual Review of Language Acquisition 2. Cognition 7: 217±283. Ramus F, Nespor M and Mehler J (1999) Correlates of Pinker S (1990) Language acquisition. In: Osherson DN linguistic rhythm in the speech signal. Cognition 73(3): (ed.) An Invitation to Cognitive Science. Volume 1: 265±292. Language. Cambridge, MA: Bradford Books/MIT Rescorla RA and Wagner AR (1972) A theory of Press. Pavlovian conditioning: variations in the effectiveness Pinker S (1994) The Language Instinct. New York: William of reinforcement and nonreinforcement. In: Black AH Morrow and Company Inc. and Prokasy WF(eds) Classical Conditioning II, Pinker S (1997) How the Mind Works. New York: WW pp. 64±99. New York: Apple-Century-Crofts. Norton and Company. Rizzolatti G and Arbib MA (1998) Language within our Pinker S and Bloom P (1990) Natural language and grasp. Trends in Neurosciences 21(5): 188±194. natural selection. Behavioral and Brain Sciences 12: [Viewpoint.] 707±784. Rizzolatti G and Arbib MA (1999) From grasping to Pinker S and Mehler J (eds) (1988) Connections and speech: imitation might provide a missing link. Trends Symbols. Cambridge, MA: MIT Press. in Neurosciences 22(4): 152. [Reply.] Posner MI (1969) Abstraction and the process of Rizzolatti G and Craighero L (1998) Spatial attention: recognition. In: Bower GH and Spence JT (eds) The mechanisms and theories. In: Sabourin M, Craik Fand Psychology of Learning and Motivation, vol. 3, pp. 44±100. Robert M (eds) Advances in Psychological Science. Volume New York: Academic Press. 2: Biological and Cognitive Aspects. Hove: Psychology Posner MI (1978) Chronometric Explorations of Mind. Press. Hillsdale: Lawrence Erlbaum Associates. Rochester N, Holland JH, Haibt LH and Duda WL (1956) Posner MI (2001) Cognitive neuroscience: the synthesis of Tests on a cell assembly theory of the action of the mind and brain. In: Dupoux E (ed.) Languge, Brain, and brain, using a large digital computer. IRE Transactions Cognitive Development: Essays in Honour of Jacques on Information Theory 2: 80±93. Mehler, pp. 403±416. Cambridge: Bradford Books/MIT Rogers H (1967/1988) Theory of Recursive Functions and Press. Effective Computability. MIT Press. [1st edn: 1967. Posner MI (ed.) (1989) Foundations of Cognitive Science. Paperback edn: 1988] Cambridge, MA: Bradford Books/MIT Press. Rosch E (1973) Natural categories. Cognitive Psychology Posner MI, Petersen SE, Fox PT and Raichle ME (1988) 7: 573±605. Localization of cognitive operations in the human Rosch E (1978) Principles of categorization. In: Rosch E brain. Science 240. and Lloyd BB (eds) Cognition and Categorization, Premack D (1972) Language in chimpanzees? Science 172: pp. 27±48. Hillsdale, NJ: Lawrence Erlbaum. 808±822. Rosch E and Mervis CB (1996) Family resemblances: Premack D (1986) Gavagai! or the Future History of the studies in the internal structure of categories. In: Animal Language Controversy. Cambridge, MA: Geirsson H and Losonsky M (eds) Readings in Language Bradford Books/MIT Press. and Mind, pp. 442±446. Cambridge: Blackwell. What is Cognitive Science? xxxix

Rosch E, Mervis CB, Gray WD and Johnson DM (1976) of a Symposium held at the National Physical Laboratory, Basic objects in natural categories. Cognitive Psychology November 1958, pp. 513±526. London: Her Majesty's 8: 382±439. Stationery Office. Rosenblatt F(1958) The perceptron: a probabilistic model Shannon CE (1938) A symbolic analysis of relay and for information storage and orgranization in the brain. switching circuits. Transactions of the American Institute Psychological Review 65: 386±408. of Electrical Engineers 57: 1±11. Rosenblatt F(1962) Principles of Neurodynamics: Shannon CE (1948) A mathematical theory of Perceptrons and the Theory of Brain Mechanisms. communication. Bell System Technical Journal Washington, DC: Spartan Books. 27: 379±423, 623±656. Rosenblueth A, Wiener N and Bigelow J (1943) Behavior, Shannon CE and Weaver W (1949/1998) The Mathematical purpose and teleology. Philosophy of Science 10: 18±24. Theory of Communication. Urbana, IL: University of Rumelhart DE and McClelland JL (1986) Parallel Illinois Press. Distributed Processing: Explorations in the Microstructure Shepard RN (1962) The analysis of proximities: of Cognition. Cambridge: MIT Press. multidimensional scaling with an unknown distance Russell B (1900) The Principles of Mathematics, vol. 1. function. Psychometrika 27(2): 125±140. Cambridge, UK: Cambridge University Press. Shepard RN (1964) On subjectively optimum selection Russell B (1919) Introduction to Mathematical Philosophy. among multiattribute alternatives. In: Bryan GL (eds) London, UK: George Allen and Unwin Ltd. Human Judgments and Optimality, p. 166. New York: Saleem KS, Pauls J, Augath MA et al. (2002) Magnetic Wiley. resonance imaging of neuronal connections in the Shepard RN (1994) Perceptual-cognitive universals as macaque monkey. Neuron 34: 685±700. reflections of the world. Psychonomic Bulletin and Samuel AL (1959) Some studies in machine learning Review 1: 2±28. using the game of checkers. IBM Journal of Research and Shepard RN and Metzler J (1971) Mental rotation of Development 3: 210±229. three-dimensional objects. Science 171: 701±703. Savage-Rumbaugh ES, Murphy J, Sevcik RA et al. Smith E, Osherson D, Rips L and Keane M (1988) (in press) Language comprehension in ape and child. Combining prototypes: a selective modification model. Monographs of the Society for Research in Child Cognitive Science 12: 485±527. Development, vol. 58. Smith EE and Medin DL (1981) Categories and Concepts. Schacter D and Tulving E (eds) (1994) Memory Systems, Cambridge, MA: Harvard University Press. Cambridge, MA: MIT Press. Smith N and Tsimpli I-M (1995) The Mind of a Savant. Scoville WB and Milner B (1957) Loss of recent memory New York and Oxford, UK: Blackwell. after bilateral hippocampal lesion. Journal of Neurology, Smith VL (2003) Experimental methods in economics. Neurosurgery, and Psychiatry 20: 11±21. Encyclopedia of Cognitive Science. London: Nature Searle JR (1980a) Intrinsic intentionality. Behavioral and Publishing Group. Brain Sciences 3: 450±456. Spelke ES (1985) Perception of unity, persistence and Searle JR (1980b) Minds, brains and programs. Behavioral identity: thoughts on infants' conception of objects. In: and Brain Sciences 3: 417±424. Mehler J and Fox R (eds) Neonate Cognition: Beyond the Searle JR (1992) The Rediscovery of the Mind. Cambridge, Blooming Buzzing Confusion. Hillsdale, NJ: Lawrence MA: MIT Press/Bradford Books. Erlbaum. Searle JR (1996) Is the brain's mind a computer program? Spelke ES (1988) Where perceiving ends and thinking In: Geirsson H and Losonsky M (eds) Readings in begins: the apprehension of objects in infancy. In: Language and Mind, pp. 264±271. New York and Yonas A (ed.) Perceptual Development in Infancy, vol. 20. Oxford, UK: Blackwell. Minneapolis, MI: University of Minnesota Press. Sebeok TA (1977) Ecumenicalism in semiotics. In: Sebeok Sperling G (1960) The information available in brief TA (ed.) APerfusion of Signs . Bloomington, IN: Indiana visual presentations. Psychological Monographs 74. University Press. Sperry RW (1968) Hemisphere deconnection and unity in Sebeok TA (1995/2000) Semiotics and the biological conscious awareness. American Psychologist 23(10): sciences: Initial conditions. [a lecture delivered at the 723±733. Collegium Budapest.] Stein D (1985) Ada: A Life and a Legacy. Cambridge, MA: Seidenberg MS and Petitto LA (1979) Signing behavior in MIT Press. apes: a critical review. Cognition 7: 177±215. Sternberg S (1966) High-speed memory scanning in Seidenberg MS and Petitto LA (1987) Communication, human memory. Science 153: 652±654. symbolic communication, and language in child and Sternberg S (1969) The discovery of processing stages: chimpanzee. Journal of Experimental Psychology extensions of Donders' method. Acta Psychologica 30: (General) 116(3): 279±287, 1483±1496. [Comment on 276±315. Savage-Rumbaugh, McDonald, Sevcik, Hopkins and Stevens T and Karmiloff-Smith A (1997) Word learning in Rupert.] a special population: do individuals with Williams Selfridge OG (1958) Pandemonium: a paradigm for syndrome obey lexical constraints? Journal of Child learning. Mechanisation of Thought Processes: Proceedings Language 24: 737±765. xl What is Cognitive Science?

Stillings NA, Feinstein MH and Garfield JL (1987) Van der Lely HKJ and Stollwerck L (1996) A grammatical Cognitive Science: An Introduction. Cambridge, MA: specific language impairment in children: an MIT Press. autosomal dominant inheritance? Brain and Language Sulloway FJ (1979) Freud, Biologist of the Mind : Beyond the 52: 484±504. Psychoanalytic Legend. New York: Basic Books. van Heijenoort J (ed.) (1967) From Frege to GoÈdel: ASource Tarski A (1935) Der Wahrheitsbegriff in den Book of Mathematical Logic, 1897±1931. Cambridge, MA: formalisierten sprachen. Studia Philosophica 1: 261±405. Harvard University Press. [German translation of a book in Polish, 1933.] Volterra V, Caselli MC, Capirci O, Vicari S and Tonucci F Tarski A (1996) The semantic conception of truth and the (2002) Early Linguistic Abilities in Italian Children with foundations of semantics. In: Martinich AP (ed.) The Williams Syndrome. [Paper presented at the Williams Philosophy of Language, 3rd edn, pp. 61±84. New York, Syndrome and the Issue of Neurogenetic Oxford: Oxford University Press. Developmental Disorders Conference, Budapest, Terrace HS, Petitto LA, Sanders RJ and Bever TG (1979) Hungary.] Can an ape create a sentence? Science 206: 891±902. Volterra V and Erting CJ (eds) (2002) From Gesture to Thaler RH (1991) Quasi-rational Economics. New York: Language in Hearing and Deaf Children. Washington DC: Russell Sage Foundation. Gallaudet University Press. Thaler RH (1992) The Winner's Curse: Paradoxes and von Neumann J (1951) The general and logical theory of Anomalies of Economic Life. New York: Russell Sage automata. In: Jeffress LA (ed.) Cerebral Mechanisms in Foundation/Free Press. Behavior: The Hixon Symposium, pp. 1±41. New York: Tolman EC (1932) Purposive Behavior in Animals and Men. John Wiley and Sons. New York: Century. von Neumann J (1957) The Computer and the Brain.New Tolman EC (1948) Cognitive maps in rats and men. Haven, CONN: Yale University Press. Psychological Review 55: 189±208. von Neumann J and Morgenstern O (1944) Theory of Tolman EC (1949) There is more than one kind of Games and Economic Behaviour. Princeton, NJ: learning. Psychological Review 56: 144±155. Princeton University Press. [1st edn: 1944. 2nd edn: Tomasello M (1999) The Cultural Origins of Human 1947.] Cognition. Cambridge, MA: Harvard University Walter WG (1953) The Living Brain. New York: Press. WW Norton. Townsend DJ and Bever TG (2001) Sentence Wanner E and Gleitman LR (eds) (1982) Language Comprehension: The Integration of Habits and Rules. Acquisition: The State of the Art. Cambridge, UK: Cambridge, MA: Bradford Books/MIT Press. Cambridge University Press. Tulving E (1972) Episodic and semantic memory. In: Watson JB (1924) Behaviorism. New York: Norton. Tulving E and Donaldson W (eds) Organization and Wernicke C (1874) Der aphasische symptomencomplex: eine Memory. New York, Academic Press. psychologische studie auf anatomischer basis. Breslau: Tulving E and Thompson D (1973) Encoding specificity Cohn und Weigert. and retrieval processes in episodic memory. Wexler K (2002) Lenneberg's dream: learning, normal Psychological Review 80: 352±373. language development and specific language Turing AM (1936a) On computable numbers: with an impairment. In: Schaffer J and Levy Y (eds) Language application to the Entscheidungsproblem (Part 1). Competence across Populations: Towards a Definition of Proceedings of the London Mathematical Society, Series 2 Specific Language Impairment, pp. 11±60. Mahwah, NJ: 42: 230±265. Erlbaum. Turing AM (1936b) On computable numbers: with an Wexler K and Borer H (1986) The Maturation of Syntax, application to the Entscheidungsproblem (Part 2). vol. 39. School of Social Sciences, University of Proceedings of the London Mathematical Society, Series 2 California, Irvine. 43: 544±546. Wexler K and Culicover PW (1980) Formal Principles of Tversky A (1977) Features of similarity. Psychological Language Acquisition. Cambridge, MA: MIT Press. Review 84: 327±352. Wexler K and Manzini MR (1986) Parameters and Tversky A and Gati I (1978) Studies of similarity. In: learnability in binding theory. In: Roeper T and Rosch E and Lloyd B (eds) Cognition and Categorization, Williams E (eds) Parameters and Linguistic Theory. pp. 79±98. Hillsdale, NJ: Lawrence Erlbaum. Whitehead AN and Russell B (1910) Principia Ungerleider LG and Mishkin M (1982) Two cortical Mathematica, vol. 1. Cambridge, UK: Cambridge visual systems. In: Ingle DJ, Goodale MA and University Press. Mandfield RJW (eds) Analysis of Visual Behavior. Wiener N (1948) Cybernetics, or Control and Communication Cambridge: MIT Press. in the Animal and the Machine. New York: John Wiley Uriagereka J (1998) Rhyme and Reason: An Introduction to and Sons. Minimalist Syntax. Cambridge, MA: MIT Press. Wildgen W (2001) Kurt Lewin and the rise of cognitive Van der Lely HKJ (1997) Language and cognitive science. In: Albertazzi L (ed.) The Dawn of Cognitive development in a grammatical SLI boy: modularity and Science: Early European Contributors, pp. 299±332. innateness. Journal of Neurolinguistics 10: 75±107. Dordrecht, Holland: Kluwer Academic. What is Cognitive Science? xli

Wilson MA and McNaughton BL (1993) Dynamics of the Winograd T (1972) Understanding Natural Language.New hippocampal ensemble code for space. Science 261: York: Academic Press. 1055±1058. Zeki SM (1978) Functional specialization in the Wilson RA and Keil FC (eds) (2000) The MIT visual cortex of the rhesus monkey. Nature 274: Encyclopedia of the Cognitive Sciences. Cambridge, MA: 423±428. MIT Press.