School of Psychology, Birkbeck College

March 22 2007

Course PSYC044U (Adaptive Learning And Comparative Cognition.)

Easter Handout for Summer-Term Lectures (2007)

gif  'pdf' version for printing

(This is a largely a re-issue of the Easter Handout from previous years: some updated material will be made available in additional paper handouts and corresponding intranet material next term. The initial pages of the handouts used in the Summer Term in 2006 are available on the intranet at http://tinyurl.com/277d25.)

Week 10 (April 26th) Primate Cognition 1 – attempts at “Language” training

Week 11 (May 3rd) Primate Cognition 2 – spatial and social knowledge and reasoning. Comparisons of human and animal learning

Week 12 (May 10th) Learning in Neural networks/Connectionism

Week 13 (May 17th) 1st lecture, Machine, animal and human learning + 2nd, Course Review

(For essays see the list dated March 15th)

Primate Cognition – Notes (Weeks 10 & 11)

Because primates are biologically more similar to people than other species, it is often assumed that cognitive processes in primates (especially in our closest living relatives, the great apes) will be noticeably human-like, and distinguishable, quantitatively or quantitatively, from those of non-primate species. It is clearly necessary to assess the behavioural evidence very carefully in order to evaluate the soundness of this assumption.

Although evidence for language-like communication in apes remains weak (despite Greenfield & Savage-Rumbaugh, 1993 and Bodamer and Gardner, 2002), experimental studies of primate cognition have implied that apes, and to a lesser extent, monkeys, are at least quantitatively superior to other species at various kinds of reasoning task (e.g. Gillan et al 1981; Menzel et al 1985).

A possible framework for comparisons of non-linguistic cognitive capacities across species is supplied by Piagetian tests designed to assess developmental changes in human cognition (cf Redshaw, 1978; Premack, 1983; Dore, 1986; ; Pasnak, 1979; Muncer, 1982; Woodruff et al, 1978).

On the other hand, there are many who disagree that primates (other than humans) have more than co-incidental cognitive advantages over other species. In particular Macphail (e.g. 1985, 1996) supports a ‘null-hypothesis’ of equal intelligence (assessed in terms of associative learning) across all vertebrate species. Others such as Herman (see Shyan and Herman, 1987) would suggest that there are species differences in cognitive ability, but that certain non-primate species (dolphins, or other large-brained mammals) demonstrate primate levels of performance on complex tasks.

Further Notes on Primate cognition and species differences in cognition

“A third hypothesis proposes that there are, in fact, neither quantitative nor qualitative differences among the intellects of non-human vertebrates. It is argued that this null hypothesis is currently to be preferred, and that man’s intellectual superiority may be due solely to our possession of a species -specific language-acquisition device” (Macphail, 1985; p.37).

“Our conclusion is that it is possible to identify two processes, use of the outcome of one trial to predict the outcome of another, and transfer of this or other rules across a change in physical stimuli, .......neither of which is uniformly distributed among vertebrates.” (Mackintosh et al, 1985; p. 62).

1. Theories

The first quotation gives Macphail’s (1985, 1987) “null hypothesis” for vertebrate intelligence – the hypothesis that any observed behavioural differences are due to “contextual variables” such as perception or motivation, which are distinguished from intellectual or cognitive mechanisms.

The second quotation gives one alternative theory – that particular mechanisms, such as the transfer of rules to different physical stimuli (as opposed to rote-learning which is closely tied to specific stimuli), may be more available to some species than others.

Material for Weeks 10 and 11 concentrates on data for primates, but it should be noted that many of these issues apply to species differences more generally. For instance Mackintosh et al (1985) are concerned with differences between pigeons (rote-learners) and corvids ( – crows: rule learners).

Theoretical positions which assume differences in cognitive mechanisms available to different vertebrate species are summarised below.

1.1 Abstraction. Learning may be tied to specific physical stimuli to a greater or lesser extent. This is one way of characterising the position of Mackintosh (1988 and et al 1985). Premack (1983) took the position that “only primates have abstract codes” and his 1986 view could be interpreted as an extension of this.

1.2 Piagetian stages. Piagetian theories of mental development can be applied to species differences – although in most cases it is development only during the human sensory-motor period which is relevant. The theoretical content (Piaget, 1971) is less important than the use of Piagetian tests of cognitive attainment, mostly connected with the attainment of different levels of object permanence.

1.3 Social skills. Although there are many other highly social vertebrates primate intelligence in particular has been related to the learning of social skills and strategies (e.g. Humphrey, 1976; Cheney and Seyfarth, 1992). This hypothesis is interesting but especially difficult to test.

1.4 “Theory of Mind” hypotheses. These variants of “social skills” idea suppose that only some species have a functional concept of “self” (Gallup, 1970) or that only particular species (usually only the great apes) are able to make inferences or assumptions about the goals and intentions of conspecifics or human experimenters (Premack and Woodruff, 1978). This has recently attract extra interest because of the hypothesis that human autism is characterised by a lack of this capacity (e.g. Leslie, 1987). The social cognitive features of “shared-reference” (or “shared-attention”) and “proto- declarative” communicative acts have been proposed as important pre-conditions both for the development of a “theory of mind” and for the development of human language (Baron-Cohen, 1992; Savage-Rumbaugh et al, 1983; Tomasello, 2000.)

2. Experimental Evidence

2.1 Learning sets, serial reversal and oddity rules. Evidence using these experimental procedures is reviewed by Mackintosh et al (1985) and Mackintosh (1988) and is particular relevant to the “Abstraction” theory. Pigeons trained on matching-to-sample (or “oddity” which involve selecting the choice different from the sample) with a particular pair of colours do not generalize to a new pair of colours, whereas corvids (which include rooks, jackdaws and blue- jays) do, and also transfer the rule to non-coloured shape stimuli. Young chimpanzees are said to sort objects on the basic of matching, or transfer the rule in a matching task, “spontaneously” (Hayes and Nissen, 1971; Oden et al, 1988)

Learning sets (Walker, 1987; pp 270-4) appear to test “learning to learn” with a very simple problem, which is to choose one of a pair of objects. Even primates will take half-a-dozen or more trials before they consistently choose the one of a pair of objects under which food is hidden. But after several hundred pairs, only one presentation is needed – there is accurate performance on the second trial (whether the random choice on the first was right or wrong). Chimpanzees and rhesus monkeys show strong learning-set performance, but rats and pigeons hardly any. However corvids tested in a similar way also show good learning sets. (e.g. Hunter and Kamil, 1971). Evidence suggests that learning-set performance is accomplished by the use of a “Win-stay, Lose-shift” strategy which is independent of particular objects, and entails remembering the outcome of trial 1 on trial 2. Successive reversal learning probably also entails the use of such a strategy, but for species such as rats and pigeons is not independent of particular physical stimuli. Goldfish show little evidence of successive reversal learning (Mackintosh et al, 1985).

2.2 Piagetian tests. Tests designed to assess the cognitive development of human infants in the first year of life can be used fairly directly for apes, and, with greater modification, for other species. Redshaw (1978) gave the same sequence of standard tests to 4 hand-reared gorilla infants and two human babies. The apes passed the vast majority of tests in the same order as babies did, a few weeks earlier. In Piagetian terms the tests were designed to measure object permanence, achieving desired events, object relations in space, and operational causality. Modified tests suggest that mammalian infants of some species also progress partially up the object permanence scale, (Doré, 1986; Gruber et al, 1971; Gagnon and Doré, 1992) with adult apes coming closest to human performance (Premack, 1983; Natale et al, 1986).

This suggests there may be differences between species in the types of cognitive processes available forming perceptual representations of objects in space. Questions may also be asked about rudimentary forms of reasoning about quantity, number and causality. Conservation-like performance has been measured in monkeys and apes (Woodruff et al, 1978; Pasnak, 1979; Muncer, 1982). Many species can learn some simple forms of number discrimination and more sophisticated assessment of number has been claimed for some specially trained chimps (Boysen and Berntson, 1989; Rumbaugh et al, 1987; Matsuzawa, 1985). Premack (1983) compares the “natural reasoning” performance of chimpanzees (requiring object permanence and the distinction between one and two objects) with that of 4- or 5- year old children.

2.3 Artificial reasoning experiments with chimpanzees. Gillan (1981) suggested that one highly trained chimp could perform transitive inference, and Gillan et al (1981) that Sarah, with 10 years of experience using plastic tokens to indicate “sameness” and “difference”, could perform well on a variety of analogical reasoning tasks (Walker, 1987; pp 344- 348).

2.4 Self-recognition, and interpretation of the behaviour of others. Naturalistic observation of groups of chimpanzees (De Waal, 1982/9; Goodall, 1991) and vervet monkeys (Cheney and Seyfarth, 1992) suggests that individual animals come to possess a rich representation of their own social relationships to others in the group, and of the intentions and perspectives of other individuals. Laboratory tests indicate that chimpanzees, but no other non-human species yet tested except orang-utans, recognise their own images in mirrors and video displays (Gallup, 1970; Menzel et al, 1985; Povinelli, 1989, Povinelli et al, 1993). It is suggested (e.g. Povinelli et al, 1990) that this ability may be related to a capacity for understanding how objects and events appear from anothers’ perspective, the “mind-reading” (Whiten and Byrne, 1988) of another animal’s (or a human experimenter’s) intentions and the ability for deception or pretence about one’s own intentions (Woodruff and Premack, 1979). However, these forms of social cognition may be interpreted as special purpose, species- specific adaptations for social organization, rather than aspects of general-purpose learning. (Cheney and Seyfarth, 1992). Some authors (e.g. Heyes, 1994, 1996, 1998) argue that there is no satisfactory evidence that primates have any special abilities in the area of social learning and imitation, or self- recognition, whereas primatologists typically believe the the cognitive capacities of the great apes can be regarded as the precursors of human cognition (Whiten et al, 1999; Tomasello and Call, 1997; Call, 2001)

 

Main Sources– Primate Cognition

Roberts, W.A (1998) Principles of Animal Cognition. Boston: McGraw-Hill. Chapter 12 “Primate Cognition”.(1 copy on short loan at BK; 4 loan copies)

Walker, S.F. (1985). Animal Thought. Routledge & Kegan Paul: London pp 305-388

Walker, S.F. (1987b). Animal Learning. Routledge & Kegan Paul: London. pp. 332-357.

Further Reading– Primate Cognition (Alternatives)

Boysen, ST and Himes, GT (1999) Current issues and emerging theories in animal cognition. Annual Review of Psychology, Vol.50, Pp.683-705.

Heyes, C.M. (1998) Theory of mind in nonhuman primates. Behavioural and Brain Sciences, 21, 108-148.

Pearce J.M. (1997) Animal Learning and Cognition 2nd Edition. Hove: Psychology Press. 156.315 PEA in new section at Birkbeck. 1 normal and 1 Short Loan copy (pages 1-14, 237-251, 253-287)

Tomasello, M. (2000). Primate cognition: Introduction to the issue. Cognitive Science, 24(3), 351-361.

Other References: Primate Cognition (Not usually necessary for further reading)

Asendorpf, J.B., and Baudonniere, P.M. (1993) Self-awareness and other-awareness - mirror self-recognition and synchronic imitation among unfamiliar peers Developmental Psychology, 29, 88-95.

Baron-Cohen, S (1992) How monkeys do things with "words". Behavioural and Brain Sciences, 15, 148-9.

Barton, RA (1996) Neocortex size and behavioral ecology in primates. Proceedings of the Royal Society of London Series B-Biological Sciences, Vol.263, No.1367, Pp.173-177.

Beran, M. J., & Washburn, D. A. (2002). Chimpanzee responding during matching to sample: Control by exclusion. Journal of the Experimental Analysis of Behavior, 78(3), 497-508.

Beran, M. J., Beran, M. M., & Menzel, C. R. (2005). Spatial memory and monitoring of hidden items through spatial displacements by chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 119(1), 14-22.

Bodamer, M. D., & Gardner, R. A. (2002). How cross-fostered chimpanzees (Pan troglodytes) initiate and maintain conversations. Journal of Comparative Psychology, 116(1), 12-26.

Boysen, S.T. and Berntson, G.G. (1989a) Numerical competence in the chimpanzee (Pan troglodytes). Journal of Comparative Psychology, 103, 23-31

Bush, E. C., & Allman, J. M. (2004). The scaling of frontal cortex in primates and carnivores. Proceedings of the National Academy of Sciences of the United States of America, 101(11), 3962-3966.

Byrne, R., & Whiten, A. (1988). Machiavellian Intelligence : Social Expertise and the Evolution of Intellect in Monkeys, Apes, and Humans. Oxford: Clarendon Press. (Senate House 67 IQD Mac)

Cheney, D.L. and Seyfarth, R.M. (1992) Précis of How monkeys see the world. Behavioural and Brain Sciences, 15, 135-182.

Crockford, C., Herbinger, I., Vigilant, L., & Boesch, C. (2004). Wild chimpanzees produce group-specific calls: a case for vocal learning? Ethology, 110(3), 221-243.

De Waal, F. (1982/89) Chimpanzee Politics. Baltimore: Johns Hopkins U. Press.

de Waal, F.B.M. and Berger, M. (2000) Payment for labour in monkeys. Nature, 404, 583.

Dore, F.Y. (1986) Object permanence in adult cats (Felix catus) Journal of Comparative Psychology, 100, 340-47.

Dumas, C. (1992) Object permanence in cats (Felis catus): An ecological approach to the study of invisible displacements. Journal of Comparative Psychology, 106, 404-10.

Elston, G. N., Benavides-Piccione, R., & DeFelipe, J. (2001). The pyramidal cell in cognition: A comparative study in human and monkey. Journal of Neuroscience, 21(17), U1-U5.

Emery, N. J., & Clayton, N. S. (2001). Effects of experience and social context on prospective caching strategies by scrub jays. Nature, 414(6862), 443-446.

Enard, W., Khaitovich, P., Klose, J., Zollner, S., Heissig, F., Giavalisco, P., Nieselt-Struwe, K., Muchmore, E., Varki, A., Ravid, R., Doxiadis, G. M., Bontrop, R. E., & Paabo, S. (2002a). Intra- and interspecific variation in primate gene expression patterns. Science, 296(5566), 340-343.

Enard, W., Przeworski, M., Fisher, S. E., Lai, C. S. L., Wiebe, V., Kitano, T., Monaco, A. P., & Paabo, S. (2002b). Molecular evolution of FOXP2, a gene involved in speech and language. Nature, 418(6900), 869-872.

Epstein, R., Lanza, R.P. and Skinner, B.F. (1981) "Self-awareness" in the pigeon. Science, 212, 694-5.

Flombaum, J. I., & Santos, L. R. (2005). Rhesus monkeys attribute perceptions to others. Current Biology, 15(5), 447-452.

Gagnon, S. & Doré, F.Y. (1992) Search behaviour of various breeds of adult dogs (Canis familiaris): Object permanence and olfactory cues. Journal of Comparative Psychology, 106, 58-68.

Gallistel, C.R. and Gelman, R. (2000) Non-verbal numerical cognition: from reals to integers. Trends in Cognitive Sciences, 4, 59-65.

Gallup, G.G., Jr. (1970) Chimpanzees: self-recognition. Science, 167, 86-7.

Gallup, GG, Povinelli, DJ, Suarez, SD, Anderson, JR, Lethmate, J and Menzel, EW (1995) Further reflections on self-recognition in primates. Animal Behaviour, Vol.50, No.Pt6, Pp.1525-1532

Gannon, PJ, Holloway, RL, Broadfield, DC and Braun, AR (1998) Asymmetry of chimpanzee planum temporale: humanlike pattern of wernicke's brain language area homolog. Science, Vol.279, No.5348, Pp.220-222.

Gillan, D.J. (1981) Reasoning in the chimpanzee: II. Transitive inference. Journal of Experimental Psychology: Animal Behaviour Processes, 7, 150-64

Gillan, D.J., Premack, D. and Woodruff, G. (1981) Reasoning in the chimpanzee: I. Analogical reasoning. Journal of Experimental Psychology: Animal Behaviour Processes, 7, 1-17.

Gisner, R. & Shusterman, R.J. (1992) Sequence, syntax, and semantics: Responses of a language-trained sea lion (Zalaphus californicus) to novel sign combinations. Journal of Comparative Psychology, 106, 78-91.

Giurfa, M., Zhang,S., Jenett,A., Menzel,R. & Srinivasan, M.V. (2001) The concepts of 'sameness' and 'difference' in an insect Nature 410, 930 - 933

Gomez, J. C. (2005). Species comparative studies and cognitive development. Trends in Cognitive Sciences, 9(3), 118-125.

Goodall, J. (1991) Through a Window: Thirty Years with the Chimpanzees of Gombe. London: Penguin.

Greenfield, P.M. and Savage-Rumbaugh, E.S. (1990) Grammatical combination in Pan paniscus: Processes of learning and invention in the evolution and development of language. In S.T. Parker and K.R. Gibson (eds) "Language" and Intelligence in Monkeys and Apes: Comparative Developmental Perspectives. Cambridge University Press, 540-578.

Greenfield, P.M. and Savage-Rumbaugh, E.S. (1993) Comparing communicative competence in child and chimp: the pragmatics of repetition. Journal of Child Language, 20, 1-26.

Gruber, H.E., Girgus, J.S. and Banuazizi, A. (1971) The development of object permanence in the cat. Developmental Psychology, 4, 9-15.

Harlow, H.F. (1949) The formation of learning sets. Psychological Review, 56, 51-65.

Hasegawa, R. P., Blitz, A. M., Geller, N. L., & Goldberg, M. E. (2000). Neurons in monkey prefrontal cortex that track past or predict future performance. Science, 290(5497), 1786-1789.

Hauser, MD, Kralik, J, Bottomahan, C, Garrett, M and Oser, J (1995) Self-recognition in primates - phylogeny and the salience of species- typical features. Proceedings of The National Academy of Sciences of The United States of America, Vol.92, No.23, Pp.10811-10814

Herman, L.M., Richards, D.G. and Wolz, J.P. (1984) Comprehension of sentences by bottlenosed dolphins. Cognition, 16, 129-219.

Heyes, CM (1994) Reflections on self-recognition in primates. Animal Behaviour, Vol.47, No.4, Pp.909-919

Heyes, CM (1996) Self-recognition in primates - irreverence, irrelevance and irony. Animal Behaviour, Vol.51, No.Pt2, Pp.470-473

Heyes, CM (1998) Theory of mind in nonhuman primates. Behavioural and Brain Sciences, 21, 108-148.

Hopkins, W. D., Marino, L., Rilling, J. K., & MacGregor, L. A. (1998). Planum temporale asymmetries in great apes as revealed by magnetic resonance imaging (MRI). Neuroreport, 9(12), 2913-2918.

Humphrey, N.K. (1976) The social function of intellect. In P.P.G. Bateson and R.A. Hinde (eds) Growing Points in Ethology. Cambridge University Press, 303-21.

Inoue-Nakamura, N, & Matsuzawa, T (1997) Development of stone tool use by wild chimpanzees (Pan troglodytes). Journal of Comparative Psychology, Vol.111, No.2, Pp.159-173.

Jensvold, M. L. A., & Gardner, R. A. (2000). Interactive use of sign language by cross-fostered chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 114(4), 335-346.

Jerison, H.J. (1973) Evolution of the Brain and Intelligence. Academic Press: London.

Kaminski, J., Call, J., & Fischer, J. (2004). Word learning in a domestic dog: Evidence for "fast mapping". Science, 304(5677), 1682-1683.

Katz, J. S., Wright, A. A., & Bachevalier, J. (2002). Mechanisms of same/different abstract-concept learning by rhesus monkeys (Macdca mulatta). Journal of Experimental Psychology-Animal Behavior Processes, 28(4), 358-368.

Kawai, N., & Matsuzawa, T. (2000). Cognition - Numerical memory span in a chimpanzee. Nature, 403(6765), 39-40.

Leslie, A.M. (1987) Pretence and representation: The origins of "theory of mind". Psychological Review, 94, 412-26.

Lilienfeld, S. O., Gershon, J., Duke, M., Marino, L., & de Waal, F. B. M. (1999). A preliminary investigation of the construct of psychopathic personality (psychopathy) in chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 113(4), 365-375.

Mackintosh, N.J., Wilson, B. and Boakes, R.A. (1985) Differences in mechanisms of intelligence among vertebrates. Philosophical Transactions of the Royal Society, B, 308, 53-65.

Macphail, E.M. (1982) Brain and Intelligence in Vertebrates. Clarendon Press: Oxford.

Macphail, E.M. (1987) The comparative psychology of intelligence. Behavioural and Brain Sciences, 10, 645-695.

Macphail, E.M. (1998) The Evolution of Consciousness. Oxford: Oxford University Press.

Macphail, EM (1996) Cognitive function in mammals - the evolutionary perspective . Cognitive Brain Research, Vol.3, No.3-4, Pp.279-290.

Maestripieri, D., Ross, S. K., & Megna, N. L. (2002). Mother-infant interactions in western lowland gorillas (Gorilla gorilla gorilla): Spatial relationships, communication, and opportunities for social learning. Journal of Comparative Psychology, 116(3), 219-227.

Mansouri, F. A., & Tanaka, K. (2002). Behavioral evidence for working memory of sensory dimension in macaque monkeys. Behavioural Brain Research, 136(2), 415-426.

Marino, L. (2002). Convergence of complex cognitive abilities in cetaceans and primates. Brain Behavior and Evolution, 59(1-2), 21-32.

Matsuzawa, T. (1985) Use of numbers by a chimpanzee. Nature, 315, 57-59.

Menzel, C. R. (1999). Unprompted recall and reporting of hidden objects by a chimpanzee (Pan troglodytes) after extended delays. Journal of Comparative Psychology, 113(4), 426-434.

Menzel, E.W., Jr., Savage-Rumbaugh, E.S. and Lawson, J. (1985) Chimpanzee ( Pan troglodytes ) spatial problem solving with the use of mirrors and televised equivalents of mirrors. Journal of Comparative Psychology, 99 , 211-217.

Muncer, S.J. (1982) 'Conservations' with a chimpanzee. Developmental Psychobiology 16, 1-11.

Muncer, S.J. and Ettlinger, G. (1981) Communication by a chimpanzee: first trial mastery of word order that is critical for meaning, but failure to negate conjunctions. Neuropsychologia, 19, 73-8.

Neel, MA (1996) Mirrors, monkeys, and group-dynamics. Primates, Vol.37, No.4, Pp.411-421.

Nieder, A. (2005). Counting on neurons: The neurobiology of numerical competence. Nature Reviews Neuroscience, 6(3), 1-14.

Oden, D.L., Thompson, R.K.R. and Premack, D. (1988) Spontaneous transfer of matching by infant chimpanzees ( Pan troglodytes ) Journal of Experimental Psychology: Animal Behaviour Processes, 14, 140-145.

Pepperberg, I.M. (1990) Cognition in an African Gray Parrot (Psittacus erithacus). Further evidence for comphrehension of categories and labels. Journal of Comparative Psychology, 104, 41-52.

Pinker, S., & Jackendoff, R. (2005). The faculty of language: what's special about it? Cognition, 95(2), 201-236.

Povinelli D.J., Rulf, A.B. and Bierschwale, D.T. (1994) Absence of knowledge attribution and self-recognition in young chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 108, 74-80.

Povinelli, D. J., Bering, J. M., & Giambrone, S. (2000). Toward a science of other minds: Escaping the argument by analogy. Cognitive Science, 24(3), 509-541.

Povinelli, D.J. (1989) Failure to find self-recognition in Asian elephants (Elephas maximus) in contrast to their use of mirror cues to discover hidden food. Journal of Comparative Psychology, 103, 122-131

Povinelli, D.J. (1993) Reconstructing the evolution of mind. American Psychologist, 48, 493-509.

Povinelli, D.J. and deBlois, S. (1992) Young children's (Homo sapiens) understanding of knowledge formation in themselves and others. Journal of Comparative Psychology, 106, 228-238.

Povinelli, D.J., Nelson, K.E. and Boysen, S.T. (1990) Inferences about guessing and knowing by chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 104, 203-210.

Povinelli, D.J., Parks, K.A., and Novak, M.A. (1991) Do rhesus monkeys (Macaca mulatta) attribute knowledge and ignorance to others? Journal of Comparative Psychology, 105, 318-325.

Povinelli, D.J., Rulf, A.B., Landau, K.R., and Bierschwale, D.T. (1993) Self-recognition in chimpanzees (Pan troglodytes): distribution, ontogeny and patterns of emergence. Journal of Comparative Psychology, 107, 347-372.

Povinelli, DJ and Preuss, TM (1995) Theory of mind - evolutionary history of a cognitive specialization. Trends in Neurosciences, Vol.18, No.9, Pp.418-424.

Premack, D. (1983). The codes of man and beasts. Behavioural and Brain Sciences, 6. 125-37.

Premack, D. (1986) Gavagai! or the Future History of the Animal Language Controversy. MIT Press: London.

Premack, D. and Premack, A.J. (1994) Levels of causal understanding in chimpanzees and young children. Cognition, 50, 346-362.

Premack, D. and Woodruff, G. (1978) Does the chimpanzee have a theory of mind? Behavioural and Brain Sciences, 4, 515-526.

Preuschoft, S. (1999). Are primates behaviorists? Formal dominance, cognition, and free-floating rationales. Journal of Comparative Psychology, 113(1), 91-95.

Redshaw, M. (1978) Cognitive development in human and gorilla infants. Journal of Human Evolution, 7, 133-41. (TIED XEROX).

Reiss, D. and McCowan, B. (1993) Spontaneous vocal mimicry and production by bottlenose dolphines (Tursiops truncatus): Evidence for vocal learning. Journal of Comparative Psychology, 107, 301-312.

Reiss, D., and Marino L. (2001) Mirror self-recognition in the bottlenose dolphin: A case of cognitive convergence. Proceedings of the National Academy of Sciences of the United States of America, 98, no 10(May 1). 5937-5942

Rendall, D, Rodman, PS, Emond, RE (1996) Vocal recognition of individuals and kin in free-ranging rhesus-monkeys. Animal Behaviour, Vol.51, No.Pt5, Pp.1007-1015.

Rizzolatti, G, Fadiga, L, Gallese, V and Fogassi, L (1996) Premotor cortex and the recognition of motor actions. Cognitive Brain Research, Vol.3, No.2, Pp.131-141.

Rumbaugh, D.M., Savage-Rumbaugh, S. and Hegel, M.T. (1987) Summation in ihe chimpanzee (Pan troglodytes) Journal of Experimental Psychology: Animal Behaviour Processes, 13, 107-115.

Savage-Rumbaugh, E.S., Pate, J.L., Lawson, J., Smith, S.T. and Rosenbaum, S. (1983) Can a chimpanzee make a statement?. Journal of Experimental Psychology: General, 112, 457-92.

Savage-Rumbaugh, S., McDonald, K, Sevcik, R.A., Hopkins, W.D. and Rupert, E (1986) Spontaneous symbol acquisition and communicative use by pigmy chimpanzees (Pan paniscus ) Journal of Experimental Psychology: General, 115, 211-235.

Semendeferi, K, Damasio, H and Frank, R (1997) The evolution of the frontal lobes: a volumetric analysis based on three-dimensional reconstructions of magnetic resonance scans of human and ape brains [full text availability]. Journal of Human Evolution, Vol.32, No.4, Pp.375-388.

Shyan, M.R. and Herman, L.M. (1987) Determinants of recognition of gestural signs in an artificial language by Atlantic bottlenosed dolphins (Tursiops truncatus) and humans (Homo sapiens) Journal of Comparative Psychology, 101, 112-125.

Slotnick, B. (2001) Animal cognition and the rat olfactory system Trends in Cognitive Sciences, 5, no 5, 216-222.

Stanford, C. B. (1998). The social behavior of chimpanzees and bonobos - empirical evidence and shifting assumptions. Current Anthropology, 39(4), 399-420.

Sulkowski, G. M., & Hauser, M. D. (2001). Can rhesus monkeys spontaneously subtract? Cognition, 79(3), 239-262.

Tomasello, M., Hare, B., & Agnetta, B. (1999). Chimpanzees, Pan troglodytes, follow gaze direction geometrically. Animal Behaviour, 58, 769-777.

Tschudin, A., Call, J., Dunbar, R. I. M., Harris, G., & van der Elst, C. (2001). Comprehension of signs by dolphins (Tursiops truncatus). Journal of Comparative Psychology, 115(1), 100-105.

Walker, S.F. and Carston, R. (1987) Multiple Review of Gavagai! or the Future History of the Animal Language Controversy (Cambridge, Mass.: MIT Press, 1986) Mind & Language, 2, 326-32.

Wallman, J. (1992) Aping Language. Cambridge University Press.

Want, S. C., & Harris, P. L. (2002). How do children ape? Applying concepts from the study of non- human primates to the developmental study of 'imitation' in children. Developmental Science, 5(1), 1-13.

Whiten, A. (2000). Primate culture and social learning. Cognitive Science, 24(3), 477-508.

Whiten, A. and Byrne, R.W. (1988) Tactical deception in primates. Behavioural and Brain Sciences, 11, 233-273.

Whiten, A., & Byrne Richard, W. (1997). Machiavellian intelligence II : extensions and evaluations. Cambridge: Cambridge University Press. (Senate House 67 IQD Mac)

Whiten, A., Goodall, J., McGrew, W. C., Nishida, T., Reynolds, V., Sugiyama, Y., Tutin, C. E. G., Wrangham, R. W., & Boesch, C. (1999). Cultures in chimpanzees. Nature, 399(6737), 682-685

Woodruff, G. and Premack, D. (1979) Intentional communication in the chimpanzee: The development of deception. Cognition, 7, 333-62.

Woodruff, G., Premack, D. and Kennel, K. (1978) Conservation of liquid and solid quantity by the chimpanzee. Science, 202, 991-4.

 

 

Animal Learning and Learning in Connectionist (Neural Network) Simulations (WEEK 12)

Notes.

(All page references are to chapters in the two volumes of “Parallel Distributed Processing edited by Rumelhart and McClelland in 1986, briefly referred to as “PDP1” and “PDP2”, except where otherwise indicated)

In the last 10 years or so a new approach to answering questions about cognitive psychology has attracted a great deal of attention. It goes under various names, including “Parallel Distributed Processing”, “PDP”, “Connectionism” and “Neural Networks”.

The “neural networks” are not real ones. The basic idea is to set up conventional computers to mimic or simulate “brain-style” information processes. The characteristics of brain-style information processing which are taken most seriously are as follows –

Neurons are slow, but numerous, and highly interconnected – they can typically operate only about 100 times per second; but since there are millions of them, each one having thousands of connections with others, it is assumed that the brain can do complicated things quickly by widespread sharing out of jobs across neurons working collectively – “parallel distributed processing”

Learning takes place when the connections between neurons are modified – an important theme is that artificial neural networks can be trained. Typically a network is connected at random to start with, but learns by somehow internalising regularities imposed on it by presenting it with a series of inputs and desired outputs.

Since there is such an emphasis on learning, and the learning is supposed to be done by neuron-like units, there are some obvious similarities with old-fashioned “Stimulus-response” theories of animal learning (Thorndike, 1949, called himself a connectionist.). But the new connectionists are mainly interested in specifically human expertise. For instance, it has been claimed that an artificial neural network connected up at random to begin with can be trained to produce the correct past-tense versions of the roots of both regular and irregular English verbs (Rumelhart and McClelland, 1986b)

[bottom of page 8 of Easter handout]

A) Similarities of new connectionism with associative theories of animal learning.

1) Learning involves modifying connection strengths – “knowledge is in the connections’

2) The currency of the system is excitation and inhibition, not symbols.

3) Learning procedures work ‘simply and incrementally’. The incremental process can be formalized mathematically, and more complex knowledge acquisition is supposed to emerge from these simpler processes. (points 1-3 all on p.132).

4) There is no central executive (p.134)

[bottom of page 8 of Easter handout]
5) Biased towards empiricism (random self- organizing networks) although there is no reason why connections could not start off innately organized. (p.139)

6) The question “Why are people smarter than rats?” is answered initially in terms of ‘more cortex’ with emphasis on the potential for “forming connections” that are useful for language.(p.143). Linguistically and culturally determined schema or “tools for thought” are entertained speculatively. (e.g Rumelhart et al, 1986, Chap 14 in PDP2)

7) The initial position minimizes serial process and specializations e.g. attention, language, the central executive.

8) Learning paradigms include single stimulus presentation, stimulus pairings, classification and regularity detection (p. 159-161).

[top of page 9 of Easter handout]
9) Learning rules are a major issue, with degree of error correction a primary question (p. 46, 54, 162). This bears some similarity to Thordikean trial and error learning, BUT there is a difference in that the major principle of instrumental learning is a change in some sort of association at the point of a successful response, where-as error-correction corresponds more to “learning from mistakes”.

10) A particular learning rule, simple association by contiguity, provides a link between old and new connectionism. McClelland et al explicitly trace this back as far as Hebb (1949) by calling their formulation the ‘Hebb rule’ (p.36) In their wording the rule is

Adjust the strength of the connection between units A and B in proportion to the product of their simultaneous activation. (p.36)

Originally it was that if an axon of neuron A repeatedly took part in firing neuron B, A’s efficiency in firing B increased (Hebb, 1949; p.62). There are versions of this in Thorndike (e.g. 1905) and many earlier sources, in particular James (1890/1985, p 534) – “When two elementary brain-processes have been active together or in immediate succession, one of them, on recurring, tends to propagate its excitement into the other. The concept of local increases in connection strength is thus an old one, but some of the best known version of old connectionism (Thorndike, Hull) strongly emphasised a supervisory influence of motivational events selectively stamping-in connections.

A further similarity here is that the basic Hebbian principle of contiguity has been rejected as the sole basis some of the most obvious empirical evidence for associations, in the case of analyses of classical conditioning, and for neoconnectionists “it turns out that the Hebb rule... has some serious limitations” (p.37), and more sophisticated rules of association formation have had to be developed for neural networks, some of which have more in common with modern theories of classical conditioning (Sutton and Barto, 1981)

11) Finally, both new and old connectionism are open to the criticism of over-simplification, and have particular difficulties in their simplest forms. It was shown that a “one layer perceptron”, a network of direct connections between a set of inputs and a set of outputs, can add little to the information in the input (Minksy and Papert, 1969; p 112-3) and thus PDP models are now multilayered and include ‘hidden units’ to remap connections between input and output. There is a rough analogy with the efforts of early stimulus-response theorists to elaborate their theories with ‘mediating responses’ ‘pure stimulus acts’ etc, which were internal sources of associations between measurable input and output

[bottom of page 9 of Easter handout]

B) Differences between neoconnectionism and associative theories of animal learning.

1) Old connectionism concentrated in the first instance on single reflex paths, whereas the essence of new connectionism is emphasis on the potential of numerous reflex paths organized in a network and operating in parallel – hence “Parallel Distributed Processing”. However, there were primitive versions of this idea in Spencer and James, in Pavlov’s “analysis and synthesis”, and in stimulus- response theory.

2) Wholistic processes (e.g. Gestalt perceptual processes) tended to be contradicted by old connectionism, but emerge more naturally in new connectionism.

3) Old connectionism was heavily involved with animal learning as a model, whereas new connectionism uses mathematical models and/or computer simulations (of parallel processes, PDP being explicitly an alternative to computer-like fast serial mechanisms)

4) Many versions of old connectionism included emphasis on motivational factors in learning rules (Thorndike, Hull) or otherwise for performance rules: New connectionism tends to ignore motivation and emotion. (but see Grossberg and Gutowski, 1987). Goals in the limited sense of error correction (e.g. the ‘delta rule’ or back-propagation; p.53) are used very often, but the bias is towards automatic learning by exposure to input or by externally imposed “teaching” that uses little that could correspond to a motivational state, or internal goal anticipation, in a neural network. It is a principle that “Learning... amounts to a very simple process that can be implemented locally at each connection without any need for overall supervision” (p.37).

5) Related to this is the fact that there is not much in current simulations that corresponds closely to instrumental learning. Hinton (1989) notes this, but suggests that it would not be difficult to implement “Reinforcement learning procedures” in which local decisions are assigned credit if they correlate with a global reinforcement signal, some work having been done on the problem of delays between the global motivational outcome and previous associations. Hinton (1989) also mentions the possibility of using “relative pay-off procedures” in which connection probabilities are altered so as to maximise expected reward.

6) Another related point is that error-reduction is the dominant focus of current connectionist learning algorithms: learning only occurs when errors are made, and compared with an imposed correct output. More traditional principles of instrumental learning suggest that learning occurs on the occasions of successful responses, and that learning can take place (and indeed may be optimal) when no errors are made at all.

7) Old connectionism became dominated by Behaviourism: an attempt was made to tie everything to explicit observable stimulus input and motor output. New connectionism is in one sense the opposite of this. Although the attempt to simulate observed human performance is one of the basic strategies these attempts are in the nature of demonstrations of the powers of neural networks, not the testing of particular models against experimental data. Similarly, old connectionists were interested in practical methods of training and therapy (e.g. Watson, Thorndike, Skinner), but this is not so far a feature of neo-connectionism.

 

Main Sources – Animal Learning and Learning in Connectionist (Neural Network) Simulations (Week 12)

Lieberman, D. (1990/1993) Learning: Behavior and Cognition. Belmont: Wadsworth. (pp. 439-455: “The Neural Network Solution: NB in the .1993 edition pp511-525.)”

Quinlan, P. (1991) Connectionism and Psychology. Harvester Wheatsheaf, Hemel Hempstead. Chapter 2 “Memory and Learning in Neural Networks” esp pp.51-56, pp.69-71, and pp. 262- 266.

Walker, S.F. (1990/1992) A brief history of connectionism and it psychological implications. AI & Society 4, 17-38. (TIED XEROX) Or Walker, S.F. (1992) A brief history of connectionism and its psychological implications. In Clark, A. and Lutz, R. (eds) Connectionism in Context. Berlin: Springer-Verlag. 123-144. (Alternative for Walker, 1990)

Further Reading – Week 12

Hinton, G.E. (1989) Connectionist learning procedures. Artificial Intelligence, 40, 185-234.

Kaplan, S. Weaver, M. and French, R.M. (1992) Active symbols and internal models: Towards a cognitive connectionism. In Clark, A. and Lutz, R. (eds) Connectionism in Context. Berlin: Springer-Verlag. 91-110. (TIED XEROX)

McClelland, J.L., Rumelhart, D.E. and Hinton, G.E. (1986) The appeal of parallel distributed processing. In Rumelhart, D.E. and McClelland, J.L. (eds) Parallel Distributed Processing. Volume 1. Foundations. London: MIT Press, 3-44.

Rumelhart, D.E. and McClelland, J.L. (1986a) PDP Models and General Issues in Cognitive Science. In Rumelhart, D.E. and McClelland, J.L. (eds) Parallel Distributed Processing. Volume 1. Foundations. London: MIT Press, 110-46

Animal and Human Learning

1. Human Cultural Learning

Critics of connectionism, and some treatments of human learning, stress the “high-level” abilities involved in systematic thought and self-regulation of learning strategies. It has recently been proposed that human learning involves unique forms of social cognition which allow for the imitation of adults by infants, the instruction of infants by adults, and collaborative learning by peers, and hence the cultural transmission of acquired knowledge (Tomasello, Kruger and Ratner, 1993; Tomasell and Rakoczy, 2003).

Summary of Tomasello et al (1993)

Tomasello et al stress that human learning is unique because it involves cultural products such as social traditions, systems of education and scientific theories which accumulate over time. This accumulation of knowledge is termed the “ratchet-effect”. They propose that cultural learning is made possible because the human species has uniquely powerful forms of social cognition which enable them to understand and take the perspective of others. They suggest that there are three stages of social- cognitive ability in human development, as in the table below.

Table 1 Major features of the three types of cultural learning.

Cultural learning process Social-Cognitive ability Concept of person Cognitive Representation

Imitative 9 months Perspective-taking (e.g. joint attention, social referencing. Intentional agents (0 order) Simple (other's perspective)

Instructed (4 years) Intersubjectivity (e.g. false-belief task, intentional deception). Mental agent (1st order) Alternating/ coordinated (other's and own perspective)

Collaborative (6 years) Recursive intersubjectivity (e.g. embedded mental-state language Reflective agent (2nd order) Integrated (dyadic intersubjectivity).

[after Tomasello et al, 1993]

Chimpanzees. One line of evidence examined by Tomasello et al (1993) is provided by chimpanzees, which can be compared with material given in weeks 12 & 13. They refer to the animals given training by human intervention as “enculturated chimpanzees”. They claim that neither wild nor enculturated chimpanzees learn collaboratively from other chimpanzees of humans, because “they do not conceive of others as reflective agents” (p. 505). Although they stress that chimpanzees cannot internalize instruction and use it to regulate their own behaviour, they are sympathetic to the view that chimpanzees enculturated by humans seem to show “sophisticated skills of perspective taking” (p.506). The argument is that chimpanzees raised with human care-givers develops the animals’ skills in joint attention, which is one of the skills involved in social cognition, and that their enhanced social cognition makes enculturated chimpanzees more capable of social learning.

2. Computer-Aided Learning (CAL)

Another area in which interest in theories of human learning has been maintained is that of computer-aided instruction or Computer-Aided Learning (CAL). Here the concept of practice by doing is still important. “Proceduralization” is another concept, and is rather like habit formation operating unconsciously. Goal structures and sub-goals are also important, so the formula “learning is a function of goal achievement and results in habits” would apply at least to some theories (e.g. Anderson, 1987a, 1987b). Immediate feedback and error minimization are practical recommendations common to traditional learning theory and some theories of learning in CAL. Clearly learning which takes place by interacting with a computer programme does not involve social cognition in the ways emphasized by Tomasello et al (1993).

The main reference here is Glaser (1990), who reviews the work of Anderson (e.g. 1987b) and other contrasting theories of human learning. Glaser’s paper reviews cognitive analyses of human performance in various domains – and computer-aided instructional programs that aim to produce competence in these domains. The sorts of domains (tasks) studied include –

· medical diagnosis

· trouble shooting of electrical circuits

· reading comprehension

· algebraic equations

· geometric proofs

· computer programming

Detailed analysis of assumptions about learning that underlie these programs indicates implicit theories of learning

Programs are for the acquisition of proceduralized knowledge, and the development of self-regulatory strategies. Glaser suggests that instructional experimentation is of increasing value to the interactive growth of learning theory and its applications.

It is pointed out that over the last 20 years, work in Cognitive Science (and cognitive psychology) has tended on the structure and organization of human performance, at the expense of the investigation of the mechanisms of learning which underlie human cognition: Newell and Simon (1972) for instance explicitly de-emphasized the study of learning.

But Glaser says “Now learning theory is re-emerging in investigations spawned by different disciplines that contribute to cognitive science. This essay examines the articulation of theoretical issues in learning that is taking place through studies of instruction that are grounded in descriptions of competence.”

A common factor in the several approaches he discusses is “an explicit cognitive task analysis; that is, the objectives of instruction are based on current knowledge of the characteristics of competent performance on a task [cf comparisons of experts with novices].

There are three major aspects of competence

A. ‘Compiled’ automatized, functional and proceduralized knowledge

B. Internalized self-regulatory control strategies.

C. The structuring of knowledge for explanation and problem solving.

{NB: A. is a common factor in all kinds of learning. B. and C. are clearly more characteristic of human than animal and machine learning)

Glaser’s Commentary (Glaser, 1990; p. 36)

Various rather different instructional programs have been discussed. They teach different categories of human performance, deal with different subject matters, and derive from different traditions. No completely general view of learning is apparent, but Glaser suggests that attempts to integrate the various approaches promise to provide a more all- encompassing theory of learning. There are several commonalities between the approaches, and one major dichotomy.

Commonalities – certain major principles are shared

§ all advocate learning in the context of working on specific problems.

§ most recommend explication of the appropriate problem-solving structure.

§ most believe that strengthening knowledge results from error-free practice.

§ but apart from knowledge compilation failure or conflict may trigger new learning.

§ Guidance and feedback is recommended

§ help with problem- solving structures cross-cuts approaches.

A dichotomy – progressive mastery or no structured transitions.

§ In the mastery approach there is a specific transition path: a progressive sequence of skills is carefully fostered.

§ In contrast, in non- structured approaches, the learners may be given tools or strategies to apply on their own: the tutor only gives assistance if it is needed.

§ However, these different methods may suit different kinds of task.

 

Main sources and further reading –Machine, Animal and Human Learning

Glaser, R. (1990) The reemergence of learning theory within instructional research. American Psychologist 45, 29-39

Hartley J. (1998) Learning and Studying: A Research Perspective. London: Routledge [378.17 HAR:Short Loan]

Howe, M.J.A.(1998) Principles of Abilities and Human Learning. Hove:Psychology Press. [153.15 HOW: Short Loan]

Tomasello, M. Kruger, A.C. and Ratner, H.H. (1993) Cultural learning. Behavioural and Brain Sciences, 16, 495-552.

 

Additional references: Both Connectionism and Animal & Human Learning

Anderson, J.R. (1987a) Methodologies for studying human knowledge. Behavioural and Brain Sciences, 10, 467-505.

Anderson, J.R. (1987b) Skill acquisition; Compilation of weak-method problem solutions. Psychological Review, 94, 192-210.

Battaglia, M., & Ogliari, A. (2005). Anxiety and panic: from human studies to animal research and back. Neuroscience and Biobehavioral Reviews, 29(1), 169-179.

Berg, T., & Schade, U. (2000). A local connectionist account of consonant harmony in child language. Cognitive Science, 24(1), 123-149.

Bingham, E. (2001). Reinforcement learning in neurofuzzy traffic signal control. European Journal of Operational Research, 131(2), 232-241.

Brickley, MR, Shepherd, JP (1996) Performance of a neural-network trained to make 3rd-molar treatment- planning decisions. Medical Decision Making, Vol.16, No.2, Pp.153-160

Carpenter, G. A. (2001). Neural-network models of learning and memory: Leading questions and an emerging framework. Trends in Cognitive Sciences, 5(3), 114-118.

Colunga, E., & Smith, L. B. (2005). From the lexicon to expectations about kinds: A role for associative learning. Psychological Review, 112(2), 347-382.

Elman, J. L. (2005). Connectionist models of cognitive development: where next? Trends in Cognitive Sciences, 9(3), 111-117.

Fletcher, J., Maybery, M. T., & Bennett, S. (2000). Implicit learning differences: A question of developmental level? Journal of Experimental Psychology-Learning Memory and Cognition, 26(1), 246-252.

Fodor, J. and Pylyshyn, Z.W. (1988) Connectionism and cognitive architecture: a critical analysis. Cognition, 28, 3-71.

Hinton, G.E. and Shallice, T. (1991) Lesioning an attractor network: investigations of acquired dyslexia. Psychological Review, 98, 74-95.

Joanisse, M. F. (2004). Specific language impairments in children - Phonology, semantics, and the English past tense. Current Directions in Psychological Science, 13(4), 156-160.

Lieberman, M. D. (2000). Intuition: A social cognitive neuroscience approach. Psychological Bulletin, 126(1), 109-137.

Mareschal, D., & Johnson, S. P. (2002). Learning to perceive object unity: a connectionist account. Developmental Science, 5(2), 151-172.

Mareschal, D., French, R. M., & Quinn, P. C. (2000). A connectionist account of asymmetric category learning in early infancy. Developmental Psychology, 36(5), 635-645.

Marslen-Wilson, W. D., & Tyler, L. K. (2003). Capturing underlying differentiation in the human language system. Trends in Cognitive Sciences, 7(2), 62-63.

McClelland, J. L., & Rogers, T. T. (2003). The parallel distributed processing approach to semantic cognition. Nature Reviews Neuroscience, 4(4), 310-322.

O'Doherty, J. P. (2004). Reward representations and reward-related learning in the human brain: insights from neuroimaging. Current Opinion in Neurobiology, 14(6), 769-776.

Page, M. (2000). Connectionist modelling in psychology: A localist manifesto. Behavioral and Brain Sciences, 23(4), 443-+.

Plaut, D.C., McClelland, J.L., Seidenberg, M.S. and Patterson, K. (1996) Understanding normal and impaired reading: computational principles in quasi-regular domains. Psychological Review, 103, 56-115.

Plunkett, K and Marchman, V (1993) From rote learning to system building - acquiring verb morphology in children and connectionist nets. Cognition, Vol.48, No.1, Pp.21-69

Plunkett, K and Sinha, C. (1992) Connectionism and developmental theory. British Journal of Developmental Psychology, 10, Pp.209-254

Plunkett, K. & Marchman, V. (1991) U-shaped learning and frequency effects in a multi-layered perceptron: Implications for child language acquisition. Cognition, 38, 43-102.

Reber, PJ, Squire, LR (1994) Parallel brain systems for learning with and without awareness. Learning & Memory, Vol.1, No.4, Pp.217-229.

Reed, J. M., & Means, L. W. (2004). Human implicit memory for irrelevant dimension values is similar to rats' incidental memory in simultaneous discrimination tasks. Behavioural Processes, 67(3), 383-393.

Rumelhart, D.E. and McClelland, J.L. (1986b) On learning the past tenses of English verbs. In McClelland, J.L. and Rumelhart, D.E (eds) Parallel Distributed Processing. Volume 2. Psychological and Biological Models. London: MIT Press, 216-271.

Rumelhart, D.E., Hinton, G.E. and McClelland, J.L. (1986) A general framework for parallel distributed processing. In Rumelhart, D.E. and McClelland, J.L. (eds) Parallel Distributed Processing. Volume 1. Foundations. London: MIT Press, 45-76.

Rumelhart, D.E., Smolensky, P., McClelland J.L. and Hinton, G.E. (1986) Schemata and sequential thought processes in PDP models. In McClelland, J.L and Rumelhart, D.E. (eds) Parallel Distributed Processing. Volume 2. Psychological and Biological Models. London: MIT Press, 7-57.

Seger, CA (1994) Implicit learning. Psychological Bulletin, Vol.115, No.2, Pp.163-196.

Shanks, DR (1995) Is human learning rational. Quarterly Journal of Experimental Psychology Section A-Human Experimental Psychology, Vol.48, No.2, Pp.257-279.

Shanks, DR and StJohn, MF (1994) Characteristics of dissociable human learning-systems. Behavioral and Brain Sciences, Vol.17, No.3, Pp.367-395.

Sutton, R.S and Barto, A.G. (1981) Toward a modern theory of adaptive networks: expectation and prediction. Psychological Review, 88, 135-171.

Thomas, M. S. C., & Mareschal, D. (2001). Metaphor as categorization: A connectionist implementation. Metaphor and Symbol, 16(1-2), 5-27.

Thorndike, E.L. (1949) Selected Writings from a Connectionist's Psychology. Greenwood Press: New York.

Tikkala, A. (2000). A connectionist word production tool for Finnish nouns with a model for vowel harmony restrictions. Computer Speech and Language, 14(1), 1-13.