The cell assembly is an important concept for cognitive psychology. Cognitive processing will to a large extent depend on the relations that can exist between different assemblies. A potential relation between assemblies can already be seen in the occurrence of (classical) conditioning. However, the resulting associations between assemblies only produce behavioristic processing or so-called regular computation. Higher-level cognitive abilities most likely result from nonregular computation. I discuss the possibility of this form of computation in terms of cell assemblies.

Theories of language production propose that utterances are constructed by a mechanism that separates linguistic content from linguistic structure, Linguistic content is retrieved from the mental lexicon, and is then inserted into slots in linguistic structures or frames. Support for this kind of model at the phonological level comes from patterns of phonological speech errors. W present an alternative account of these patterns using a connectionist or parallel distributed proceesing (PDP) model that learns to produce sequences of phonological features. The (...) model's errors exhibit some of the properties of human speech errors, specifically, properties that have been attributed to the action of phonological rules, frames, or other structural generalizations. (shrink)

Consciousness and experience, William G. Lycan. Cambridge, MA: MIT Press, 1996. ISBN 0–262–12197–2 (hc)Mind as motion: explorations in the dynamics of cognition, Robert Port & Timothy Van Gelder (Eds). Cambridge, MA: MIT/Bradford, 1995. ISBN 0–262–16150–8Perception, Kathleen Akins (Ed.). New York: Oxford University Press, 1996. ISBN 0–19–508461 (hbk); 0–19–508462–4 (pbk)Context and consciousness, B. Nardi (Ed.). Cambridge, MA: MIT Press, 1996. ISBN: 0–262–14058–6Catching ourselves in the act, Horst Hendriks‐Jansen. Cambridge, MA: MIT Press, 1996. ISBN 0–262–08246–2The search for mind. A new foundation for (...) cognitive science, Sean O'Nuallain. Norwood, NJ: Ablex, 1995Mindblindness: an essay on autism and theory of mind, Simon Baron‐Cohen. Cambridge, MA: MIT Press, 1995. ISBN 0–262–02384–9. (shrink)

There is widespread belief that connectionist networks are dramatically different from classical or symbolic models. However, connectionists rarely test this belief by interpreting the internal structure of their nets. A new approach to interpreting networks was recently introduced by Berkeley et al. (1995). The current paper examines two implications of applying this method: (1) that the internal structure of a connectionist network can have a very classical appearance, and (2) that this interpretation can provide a cognitive theory that cannot be (...) dismissed as a mere implementation. (shrink)

Evidence from aphasia is considered that leads to a distinction between abstract and concrete thought processes and hence for a distinction between rules and similarity. It is argued that perceptual classification is inherently a rule-following procedure and these rules are unable to be followed when a patient has difficulty with name comprehension and retrieval.

In this paper, I define tacit knowledge as a kind of causal-explanatory structure, mirroring the derivational structure in the theory that is tacitly known. On this definition, tacit knowledge does not have to be explicitly represented. I then take the notion of a modular theory, and project the idea of modularity to several different levels of description: in particular, to the processing level and the neurophysiological level. The fundamental description of a connectionist network lies at a level between the processing (...) level and the physiological level. At this level, connectionism involves a characteristic departure from modularity, and a correlative absence of syntactic structure. This is linked to the fact that tacit knowledge descriptions of networks are only approximately true. A consequence is that strict causal systematicity in cognitive processes poses a problem for the connectionist programme. (shrink)

The cell assembly as a simple attractor cannot explain many cognitive phenomena. It must be a highly structured network that can sustain highly structured excitation patterns. Moreover, a cell assembly must be more widely distributed in space than on a square millimeter.

The study of signed languages has inspired scientific' speculation regarding foundations of human language. Relationships between the acquisition of sign language in apes and man are discounted on logical grounds. Evidence from the differential hreakdown of sign language and manual pantomime places limits on the degree of overlap between language and nonlanguage motor systems. Evidence from functional magnetic resonance imaging reveals neural areas of convergence and divergence underlying signed and spoken languages.

I enlarge on the theme that the brain mechanisms required for languageand other aspects of the human mind evolved through selective changes in the regulatory genes governing growth. Extension of the period of postnatal growth increases the role of the environment in structuring the brain, and spatiotemporal programming (heterochrony) ofgrowth might explain hierarchical representation, hemispheric specialization, and perhaps sex differences.

Marinov''s critique I argue, is vitiated by its failure to recognize the distinctive role of superposition within the distributed connectionist paradigm. The use of so-called subsymbolic distributed encodings alone is not, I agree, enough to justify treating distributed connectionism as a distinctive approach. It has always been clear that microfeatural decomposition is both possible and actual within the confines of recognizably classical approaches. When such approaches also involve statistically-driven learning algorithms — as in the case of ID3 — the fundamental (...) differences become even harder to spot. To see them, it is necessary to consider not just the nature of an acquired input-output function but the nature of the representational scheme underlying it. Differences between such schemes make themselves best felt outside the domain of immediate problem solving. It is in the more extended contexts of performance DURING learning and cognitive change as a result of SUBSEQUENT training on new tasks (or simultaneous training on several tasks) that the effects of superpositional storage techniques come to the fore. I conclude that subsymbols, distribution and statistically driven learning alone are indeed not of the essence. But connectionism is not just about subsymbols and distribution. It is about the generation of whole subsymbol SYSTEMS in which multiple distributed representations are created and superposed. (shrink)

Connectionism and classicism, it generally appears, have at least this much in common: both place some notion of internal representation at the heart of a scientific study of mind. In recent years, however, a much more radical view has gained increasing popularity. This view calls into question the commitment to internal representation itself. More strikingly still, this new wave of anti-representationalism is rooted not in armchair theorizing but in practical attempts to model and understand intelligent, adaptive behavior. In this paper (...) we first present, and then critically assess, a variety of recent anti-representationalist treatments. We suggest that so far, at least, the sceptical rhetoric outpaces both evidence and argument. Some probable causes of this premature scepticism are isolated. Nonetheless, the anti-representationalist challenge is shown to be both important and progressive insofar as it forces us to see beyond the bare representational/non-representational dichotomy and to recognize instead a rich continuum of degrees and types of representationality. (shrink)

Ramsey, Stich and Garon's recent paper Connectionism, Eliminativism, and the Future of Folk Psychology claims a certain style of connectionism to be the final nail in the coffin of folk psychology. I argue that their paper fails to show this, and that the style of connectionism they illustrate can in fact supplement, rather than compete with, the claims of a theory of cognition based in folk psychology's ontology. Ramsey, Stich and Garon's argument relies on the lack of easily identifiable symbols (...) inside the connectionist network they discuss, and they suggest that the existence of a system which behaves in a cognitively interesting way, but which cannot be explained by appeal to internal symbol processing, falsifies central assumptions of folk psychology. My claim is that this argument is flawed, and that the theorist need not discard folk psychology in order to accept that the network illustrated exhibits cognitively interesting behaviour, even if it is conceded that symbols cannot be readily identified within the network. (shrink)

This paper refines a controversial proposal: that core systems belong to a perceptual kind, marked out by the format of its representational outputs. Following Susan Carey, this proposal has been understood in terms of core representations having an iconic format, like certain paradigmatically perceptual outputs. I argue that they don’t, but suggest that the proposal may be better formulated in terms of a broader analogue format type. Formulated in this way, the proposal accommodates the existence of genuine icons in perception, (...) and avoids otherwise troubling objections. (shrink)

Abstract The paper considers two recent accounts of the difference between human and animal thought. One deflationary account, due to Daniel Dennett, insists that the only real difference lies in our ability to use words and sentences to give artificial precision and determinacy to our mental contents. The other, due to Paul Smolensky, conjectures that we at times deploy a special purpose device (the Conscious Rule Interpreter) whose task is to deal with public, symbolically coded data and commands. Both these (...) accounts make a crucial error. They offer what is in effect an extra top?level processor to soothe our realist/classical prejudices. But in each case the extra ingredient turns out to be explanatorily hollow. Appealing to language use and language processing alone mistakes a cognitive effect for a cognitive cause. I argue instead that we need to seek a more profound architectural condition which may ground our conscious linguistic abilities but also explains a variety of deeper facts. I sketch a picture which seems to meet those needs and draw out its implications for the debates about belief and about classical Artificial Intelligence. (shrink)

There is a school of thought which links connectionist models of cognition to eliminativism-the thesis that the constructs of commonsense psychology do not exist. This way of construing the impact of connectionist modelling is, I argue, deeply mistaken and depends crucially on a shallow analysis of the notion of explanation. I argue that good, higher level descriptions may group together physically heterogenous mechanisms, and that the constructs of folk psychology may fulfil such a grouping function even if they fail to (...) pick out discrete states of individual processing. More speculatively, the paper goes on to suggest that the virtues which recommend such constructs from a theorist’s third person perspective may nonetheless have close analogues in individual processing. Various difficulties for what I term a pure distributed connectionism may be solved by systems which utilise both distributed and classical symbolic representations-the latter exhibiting the discreteness and semantic interpretability which the eliminativist requires to vindicate common-sense psychology. If human beings turned out to be such mixed systems, then the eliminativist claim would be doubly misguided. It would be false as a conditional since even if pure distributed connectionism were a complete and accurate formal model of individual processing, it would not follow that the other, higher level constructs were not accurate and essential tools for a different kind of explanation. And it would involve a false antecedent since in a mixed system the symbolic descriptions may indeed be incarnate in the system’s own individual processing. The paper is in four main sections. Section 1 attempts to systematise a certain received picture of the relative status and accuracy of various levels of description of a connectionist system. Section 2 then introduces a general model of explanations which aim to group systems into equivalence classes defined for various purposes. Each grouping requires a special vocabulary and the constructs of any given vocabulary are legitimate just insofar as the grouping is interesting and useful. Section 3 goes on to show that, relative to such a model of explanation, the constructs of both symbolic AI and commonsense psychology may have a legitimate role to play in giving psychological explanations. This role is not just that of a useful approximation. The paper ends with a speculative section in which the argument for the theoretical usefulness of such symbolic constructs is extended, in a very natural way, to the domain of individual processing. Here the cognizer, in the process of regulating, debugging and understanding her own representations, creates symbols to stand for sets of distributed activity patterns. The section points out some difficulties for a pure distributed approach which may be eased by the addition of such symbolic constructs and relates the speculations to the current debate over the ’correct’architecture of cognition. If the speculations are on target, this whole debate turns out to be fundamentally illposed. (shrink)

Memory is fleeting. New material rapidly obliterates previous material. How, then, can the brain deal successfully with the continual deluge of linguistic input? We argue that, to deal with this “Now-or-Never” bottleneck, the brain must compress and recode linguistic input as rapidly as possible. This observation has strong implications for the nature of language processing: (1) the language system must “eagerly” recode and compress linguistic input; (2) as the bottleneck recurs at each new representational level, the language system must build (...) a multilevel linguistic representation; and (3) the language system must deploy all available information predictively to ensure that local linguistic ambiguities are dealt with “Right-First-Time”; once the original input is lost, there is no way for the language system to recover. This is “Chunk-and-Pass” processing. Similarly, language learning must also occur in the here and now, which implies that language acquisition is learning to process, rather than inducing, a grammar. Moreover, this perspective provides a cognitive foundation for grammaticalization and other aspects of language change. Chunk-and-Pass processing also helps explain a variety of core properties of language, including its multilevel representational structure and duality of patterning. This approach promises to create a direct relationship between psycholinguistics and linguistic theory. More generally, we outline a framework within which to integrate often disconnected inquiries into language processing, language acquisition, and language change and evolution. (shrink)

Amit's work addresses a critical issue in cognitive science: the structure of neural representations. The use of Hebbian cell assemblies is a positive step, and we now need to consider its role in a larger cognitive theory. When considering the dynamics of a system built out of attractors, a more limited version of reverberation becomes necessary.

The debate over how best to characterize inflectional morphology has been couched largely in terms of the “dual-mechanism” approach described in Pinker (Words and rules: the ingredients of language, Basic Books, 1999) versus “single-mechanism” connectionist approaches derived from Rumelhart and McClelland (On learning past tenses of English verbs, MIT, 1986). There are, however, other single-mechanism approaches. The exemplar-based or analogical models of Daelemans et al. (TimBL: Tilburg Memory-Based Learner, version 4.3 reference guide, ILK, 2002) and Skousen (Analogical modeling of language, (...) Kluwer Academic, 1989) also model inflectional usage accurately within a single-mechanism. The most striking theoretical claim peculiar to these purely analogical models is that they do not posit any resident linguistic generalizations for processing language. Instead they process new instances of usage by comparing them systematically to remembered instances of previous usage. Based on a comparison of their Minimal Generalization Model with an adaptation of Nosofsky's (Journal of Mathematical Psychology 34: 393–418, 1990) Generalized Context Model, Albright and Hayes (Cognition 90: 119–161, 2003) argue that such purely analogical models are intrinsically inadequate for modeling the English past tense. This paper shows, however, that Skousen's (Analogical modeling of language, Kluwer Academic, 1989) Analogical Model performs as well as the Minimal Generalization Model. The implications of these results for cognitive linguistics are discussed. (shrink)

The ability to combine words into novel sentences has been used to argue that humans have symbolic language production abilities. Critiques of connectionist models of language often center on the inability of these models to generalize symbolically (Fodor & Pylyshyn, 1988; Marcus, 1998). To address these issues, a connectionist model of sentence production was developed. The model had variables (role‐concept bindings) that were inspired by spatial representations (Landau & Jackendoff, 1993). In order to take advantage of these variables, a novel (...) dual‐pathway architecture with event semantics is proposed and shown to be better at symbolic generalization than several variants. This architecture has one pathway for mapping message content to words and a separate pathway that enforces sequencing constraints. Analysis of the model's hidden units demonstrated that the model learned different types of information in each pathway, and that the model's compositional behavior arose from the combination of these two pathways. The model's ability to balance symbolic and statistical behavior in syntax acquisition and to model aphasic double dissociations provided independent support for the dual‐pathway architecture. (shrink)

A U-shaped curve in a cognitive-developmental trajectory refers to a three-step process: good performance followed by bad performance followed by good performance once again. U-shaped curves have been observed in a wide variety of cognitive-developmental and learning contexts. U-shaped learning seems to contradict the idea that learning is a monotonic, cumulative process and thus constitutes a challenge for competing theories of cognitive development and learning. U-shaped behavior in language learning (in particular in learning English past tense) has become a central (...) topic in the Cognitive Science debate about learning models. Antagonist models (e.g., connectionism versus nativism) are often judged on their ability of modeling or accounting for U-shaped behavior. The prior literature is mostly occupied with explaining how U-shaped behavior occurs. Instead, we are interested in the necessity of this kind of apparently inefficient strategy. We present and discuss a body of results in the abstract mathematical setting of (extensions of) Gold-style computational learning theory addressing a mathematically precise version of the following question: Are there learning tasks that require U-shaped behavior? All notions considered are learning in the limit from positive data. We present results about the necessity of U-shaped learning in classical models of learning as well as in models with bounds on the memory of the learner. The pattern emerges that, for parameterized, cognitively relevant learning criteria, beyond very few initial parameter values, U-shapes are necessary for full learning power! We discuss the possible relevance of the above results for the Cognitive Science debate about learning models as well as directions for future research. (shrink)

Müller argues that double dissociations do not imply underlying modularity of the cognitive system, citing neural networks as examples of fully distributed systems that can give rise to double dissociations. We challenge this claim, noting that suchdouble dissociations typically do not “scale-up,” and that even some singledissociations can be difficult to account for in a distributed system.

A fundamental claim associated with parallel distributed processing theories of cognition is that knowledge is coded in a distributed manner in mind and brain. This approach rejects the claim that knowledge is coded in a localist fashion, with words, objects, and simple concepts, that is, coded with their own dedicated representations. One of the putative advantages of this approach is that the theories are biologically plausible. Indeed, advocates of the PDP approach often highlight the close parallels between distributed representations learned (...) in connectionist models and neural coding in brain and often dismiss localist theories as biologically implausible. The author reviews a range a data that strongly challenge this claim and shows that localist models provide a better account of single-cell recording studies. The author also contrast local and alternative distributed coding schemes and argues that common rejection of grandmother cell theories in neuroscience is due to a misunderstanding about how localist models behave. The author concludes that the localist representations embedded in theories of perception and cognition are consistent with neuroscience; biology only calls into question the distributed representations often learned in PDP models. (shrink)

Deep neural networks (DNNs) have had extraordinary successes in classifying photographic images of objects and are often described as the best models of biological vision. This conclusion is largely based on three sets of findings: (1) DNNs are more accurate than any other model in classifying images taken from various datasets, (2) DNNs do the best job in predicting the pattern of human errors in classifying objects taken from various behavioral datasets, and (3) DNNs do the best job in predicting (...) brain signals in response to images taken from various brain datasets (e.g., single cell responses or fMRI data). However, these behavioral and brain datasets do not test hypotheses regarding what features are contributing to good predictions and we show that the predictions may be mediated by DNNs that share little overlap with biological vision. More problematically, we show that DNNs account for almost no results from psychological research. This contradicts the common claim that DNNs are good, let alone the best, models of human object recognition. We argue that theorists interested in developing biologically plausible models of human vision need to direct their attention to explaining psychological findings. More generally, theorists need to build models that explain the results of experiments that manipulate independent variables designed to test hypotheses rather than compete on making the best predictions. We conclude by briefly summarizing various promising modeling approaches that focus on psychological data. (shrink)

Clahsen's claim that output forms of productive processes are never listed in the lexicon is a consequence of the rule/list fallacy, empirically incorrect, and not necessary for the hypothesis that the human language faculty has a dual structure, that is, a lexicon and a set of rules.

There is a gap between two different modes of computation: the symbolic mode and the subsymbolic (neuron-like) mode. The aim of this paper is to overcome this gap by viewing symbolism as a high-level description of the properties of (a class of) neural networks. Combining methods of algebraic semantics and non-monotonic logic, the possibility of integrating both modes of viewing cognition is demonstrated. The main results are (a) that certain activities of connectionist networks can be interpreted as non-monotonic inferences, and (...) (b) that there is a strict correspondence between the coding of knowledge in Hopfield networks and the knowledge representation in weight-annotated Poole systems. These results show the usefulness of non-monotonic logic as a descriptive and analytic tool for analyzing emerging properties of connectionist networks. Assuming an exponential development of the weight function, the present account relates to optimality theory – a general framework that aims to integrate insights from symbolism and connectionism. The paper concludes with some speculations about extending the present ideas. (shrink)

Interpreting the Miyashita et al. experiments in terms of a cellassembly representation does not adequately explain the performance of Miyashita's monkeys on novel stimuli. We will argue that the latter observations point to acompositionalrepresentation and suggest a dynamics involving rapid and reversible binding of distinct activity patterns.

Müller misconstrues autonomy to mean strict locality of brain function, something quite different from the functional autonomy that linguists claim. Similarly, he misperceives the interaction of learned and innate components hypothesized in current generative models. Evidence from sign languages, Creole languages, and neurological studies of rare forms of aphasia also argues against his conclusions.