Connectionism and the problem of systematicity: why Smolensky's solution doesn't work

In two recent papers, Paul Smolensky responds to a challenge Jerry Fodor and Zenon Pylyshyn posed for connectionist theories of cognition: to explain the existence of systematic relations among cognitive capacities without assuming that mental processes are causally sensitive to the constituent structure of mental representations. Smolensky thinks connectionists can explain systematicity if they avail themselves of "distributed" mental representation. In facts, Smolensky offers two accounts of distributed mental representation, corresponding to his notions of "weak" and "strong" compositional structure. We argue that weak compositional structure is irrelevant to the systematicity problem and of dubious internal coherence. We then argue that strong compositional (tensor product) representations fail to explain systematicity because they fail to exhibit the sort of constituents that can provide domains for structure sensitive mental processes.     

Systematicity redux

One of the main challenges that Jerry Fodor and Zenon Pylyshyn (Cognition 28:3–71, 1988) posed for any connectionist theory of cognitive architecture is to explain the systematicity of thought without implementing a Language of Thought (LOT) architecture. The systematicity challenge presents a dilemma: if connectionism cannot explain the systematicity of thought, then it fails to offer an adequate theory of cognitive architecture; and if it explains the systematicity of thought by implementing a LOT architecture, then it fails to offer an alternative to the LOT hypothesis. Given that thought is systematic, connectionism can offer an adequate alternative to the LOT hypothesis only if it can meet the challenge. Although some critics tried to meet the challenge, others argued that it need not be met since thought is not in fact systematic; and some claimed not to even understand the claim that thought is systematic. I do not here examine attempts to answer the challenge. Instead, I defend the challenge itself by explicating the notion of systematicity in a way that I hope makes clear that thought is indeed systematic, and so that to offer an adequate alternative to the LOT hypothesis, connectionism must meet the challenge.     

Fodor’s riddle of abduction

How can abductive reasoning be physical, feasible, and reliable? This is Fodor’s riddle of abduction, and its apparent intractability is the cause of Fodor’s recent pessimism regarding the prospects for cognitive science. I argue that this riddle can be solved if we augment the computational theory of mind to allow for non-computational mental processes, such as those posited by classical associationists and contemporary connectionists. The resulting hybrid theory appeals to computational mechanisms to explain the semantic coherence of inference and associative mechanisms to explain the efficient retrieval of relevant information from memory. The interaction of these mechanisms explains how abduction can be physical, feasible, and reliable.     

Cognition without classical architecture

Fodor and Pylyshyn (1988) argue that any successful model of cognition must use classical architecture; it must depend upon rule-based processing sensitive to constituent structure. This claim is central to their defense of classical AI against the recent enthusiasm for connectionism. Connectionist nets, they contend, may serve as theories of the implementation of cognition, but never as proper theories of psychology. Connectionist models are doomed to describing the brain at the wrong level, leaving the classical view to account for the mind.This paper considers whether recent results in connectionist research weigh against Fodor and Pylyshyn's thesis. The investigation will force us to develop criteria for determining exactly when a net is capable of systematic processing. Fodor and Pylyshyn clearly intend their thesis to affect the course of research in psychology. I will argue that when systematicity is defined in a way that makes the thesis relevant in this way, the thesis is challenged by recent progress in connectionism.     

Connectionist semantic systematicity

Fodor and Pylyshyn [Fodor, J. A., & Pylyshyn, Z. W. (1988). Connectionism and cognitive architecture: A critical analysis. Cognition, 28, 3–71] argue that connectionist models are not able to display systematicity other than by implementing a classical symbol system. This claim entails that connectionism cannot compete with the classical approach as an alternative architectural framework for human cognition. We present a connectionist model of sentence comprehension that does not implement a symbol system yet behaves systematically. It consists in a recurrent neural network that maps sentences describing situations in a microworld, onto representations of these situations. After being trained on particular sentence–situation pairs, the model can comprehend new sentences, even if these describe new situations. We argue that this systematicity arises robustly and in a psychologically plausible manner because it depends on structure inherent in the world.     

Can Compositionality Solve the Thought-or-Language Problem?

Abstract

Jerry Fodor has claimed to have a solution to the traditional problem of what comes first, thought or language. Compositionality, he says, will give us the answer, for at least one must be compositional, and if only one of them is, that is the one that has underived semantic content. He argues that natural languages are not compositional, and therefore that the content of language is derived from the content of thought. I will argue that the idea that language is not compositional conflicts with his productivity and systematicity arguments for the existence of a language of thought. I will also show that Fodor’s solution to the problem fails, as his main argument is circular. Finally, I suggest that Fodor’s argument against the compositionality of language is not decisive, and that we can still attribute at least some degree of compositionality to language.     

Two Ways to Smoke a Cigarette

In the early part of the paper, I attempt to explain a dispute between two parties who endorse the compositionality of language but disagree about its implications: Paul Horwich, and Jerry Fodor and Ernest Lepore. In the remainder of the paper, I challenge the thesis on which they are agreed, that compositionality can be taken for granted. I suggest that it is not clear what compositionality involves nor whether it obtains. I consider some kinds of apparent counterexamples, and compositionalist responses to them in terms of covert indexicality and unspecific meanings. I argue that the last option is the best for most of the cases I consider. I conclude by stressing, as against Horwich and Fodor and Lepore, that the appropriate question concerns the extent to which compositionality obtains in a natural language, rather than whether it obtains or not, so that the answer is essentially messy, requiring detailed consideration of a wide range of examples.     

Reasons explanations (of actions) as structural explanations

Non-causal accounts of action explanation have long been criticized for lacking a positive thesis, relying primarily on negative arguments to undercut the standard Causal Theory of Action (Wilson and Shpall , in: Zalta (ed) The Stanford Encyclopedia of Philosophy, 2016). Additionally, it is commonly thought that non-causal accounts fail to provide an answer to Donald Davidson’s (1963) challenge for theories of reasons explanations of actions. According to Davidson’s challenge, a plausible non-causal account of reasons explanations must provide a way of connecting an agent’s reasons, not only to what she ought to do, but to what she actually does. That is, such explanations must be truth-apt, not mere rationalizations. My aim in this paper is to show how a non-causal account of action can provide explanations that are truth-apt and genuinely explanatory. To make this argument, I take as a given an account of the practical syllogism (the syllogistic form of practical reasoning) discussed by Michael Thompson (Life and action: elementary structures of practice and practical thought, Harvard University Press, Cambridge, 2008) and Eric Wiland (Reasons, Continuum, New York, 2012), according to which the practical syllogism is truly practical rather than propositional in nature. Next, I present my primary positive thesis: reasons for actions have explanatory power in virtue of being parts of a structure—the practical syllogism—that contains the action being explained. I then argue that structural action explanations can meet Davidson’s challenge and that they genuinely explain actions. Finally, I conclude by addressing some objections to my argument.

     

Jerry Fodor on non-conceptual content

Proponents of non-conceptual content have recruited it for various philosophical jobs. Some epistemologists have suggested that it may play the role of “the given” that Sellars is supposed to have exorcised from philosophy. Some philosophers of mind (e.g., Dretske) have suggested that it plays an important role in the project of naturalizing semantics as a kind of halfway between merely information bearing and possessing conceptual content. Here I will focus on a recent proposal by Jerry Fodor. In a recent paper he characterizes non-conceptual content in a particular way and argues that it is plausible that it plays an explanatory role in accounting for certain auditory and visual phenomena. So he thinks that there is reason to believe that there is non-conceptual content. On the other hand, Fodor thinks that non-conceptual content has a limited role. It occurs only in the very early stages of perceptual processing prior to conscious awareness. My paper is examines Fodor’s characterization of non-conceptual content and his claims for its explanatory importance. I also discuss if Fodor has made a case for limiting non-conceptual content to non-conscious, sub-personal mental states. This paper has grown out of comments I made on Fodor’s paper “Revenge of the Given,” delivered at The Steven Humphrey Excellence in Philosophy Conference: “Content and Concepts: A Conference on the Philosophy of Mind” at the University of California, Santa Barbara, February 14, 2004. Thanks to Jerry Fodor for useful feedback, and to Barry Loewer for discussing with me the ideas that went into this paper.     

A Poetics of Psychological Explanation

Intentional, 'commonsense,' or 'folk' psychology is, as Jerry Fodor has remarked, ubiquitous. Explanations of what we say and do in terms of our reasons for acting are the stock in trade of intentional psychology. But there is a question whether explanations in terms of reasons are properly explanatory. Donald Davidson and Daniel Dennett, to name two, have defended intentional psychology and its reason-explanations. Still, many philosophers – including Fodor, Davidson and Dennett – fail to pay due attention to the narrative basis of such agent-centered accounts of action. In this paper, I argue that psychological explanation is an agent-centered, narrative-based interpretive practice. To make my case, I present a poetics of psychological explanation: seven elements which collectively describe what makes psychological explanations work. Narrative form allows us to represent the temporal arc of agents' actions – as well as the temporal arc of their reasoning about their actions, both prospective and retrospective. It allows us to negotiate between the canonical and the exceptional in human experience, and thus to account for actions that strike us as puzzling or unusual – whether the puzzle originates in our suboptimal understanding or the agent's suboptimal reasoning. And it allows us to juxtapose different perspectives on any action. Such juxtapositioning gives us a mechanism for coming to see how an action that strikes us as misguided might have been construed by the agent as reasonable given her understanding of her circumstances. After establishing the seven elements of the poetics, I address the objection that narrative-based accounts of intentional action are not properly explanatory.     

A Non-Compositional Inferential Role Theory

I propose a version of inferential role theory which says that having a concept is having the disposition to draw most of the inferences based on the stereotypical features associated with this concept. I defend this view against Fodor and Lepore’s objection that it violates compositionality. I show that it is possible to account for productivity and systematicity without assuming compositionality.     

Is Natural Selection in Trouble? When Emotions Run High in a Philosophical Debate

This paper deals in detail with a fairly recent philosophical debate centered around the ability of the theory of natural selection to account for those phenotypical changes which can be argued to make organisms better adapted to their environments. The philosopher and cognitive scientist Jerry Fodor started the debate by claiming that natural selection cannot do the job. He follows two main lines of argumentation. One is based on an alleged conceptual defect in the theory, the other on alleged empirical problems in it as well as empirical alternatives to it. Four philosophers and two biologists respond in a way that displays what might easily be described as fallacious. The paper relies on the ideal model of critical discussion of pragma-dialectics to offer a step-by-step analysis of the whole debate, which extended for four issues of the London Review of Books, from October 2007 through January 2008. This pragma-dialectical analysis is carried out by constant reference to the various questions (problems, issues) that arise in the debate. The analysis includes as much detail as possible both in Fodor’s original argument and in the critics’ various comments as well as Fodor’s replies along two rounds of debate. Since a simple negative evaluation in terms of fallacies is out of the question in view of the proved argumentative accomplishments of the participants, an alternative explanation is offered: the undeniable derailments in strategic maneuvering are due to the fact that, whilst ostensibly discussing the theory of natural selection, Fodor’s detractors are worried by an underlying issue, namely, the dangers of discussing the merits and demerits of natural selection as a theory of evolution in a venue as exposed to the general public as the London Review of Books, given the religiously inspired movements that threaten the teaching of evolutionary biology in schools.

     

Connectionist representation,semantic compositionality,and the instability of concept structure

Summary In the present paper connectionist approaches to the problem of internal representation and the nature of concepts are discussed. In the first part the concept of representation that underlies connectionist modeling is made explicit. It is argued that the connectionist view of representation relies on a correlational theory of semantic content- i.e., the covariation between internal and external states is taken as the basis for ascribing meaning to internal states. The problems and virtues of such a correlational approach to internal representation are addressed. The second part of the paper is concerned with whether connectionism is capable of accounting for the apparent productivity and systematicity of language and thought. There is an evaluation of the recent arguments of Fodor and Pylyshyn, who claim that systematicity can only be explained if one conceives of mental representations as structured symbols composed of context-free constituents. There is a review of empirical evidence that strongly suggests that concepts are not fixed memory structures and that the meaning of constituent symbols varies, depending on the context in which they are embedded. On the basis of this review it is concluded that the meaning of a complex expression is not computed from the context-free meanings of the constituents, and that strong compositionality, as endorsed by Fodor and Pylyshyn (1988), seems implausible as a process theory for the comprehension of complex concepts. Instead, the hypothesis is endorsed that constraint satisfaction in distributed connectionist networks may allow for an alternative account of weak compositionality compatible with the context sensitivity of meaning. In the final section, it is argued that neither mere implementation of a language of thought in connectionist networks nor radical elimination of symbol systems seems to be a fruitful research strategy, but that it might be more useful to discuss how connectionist systems can develop the capacity to use external symbol systems like language or logic without instantiating symbol systems themselves.     

The sex difference in self-assessed fears

Although the greater self-attribution of fear among females than males is well established, limited empirical information is available to explain this difference. This study assessed the contribution of Adjective Check List measures of masculinity and femininity plus the Marlowe-Crowne Social Desirability Scale to the prediction of self-ratings of fear. Significant sex differences were found for self-rated fears and for masculinity and femininity scales of the Adjective Check List. In separate analyses for males and females only the femininity measure was significantly correlated with fear scores. For males, the multiple correlation of the independent variables with the fear measure accounted for approximately 10% of the variance, while for females approximately 28% of the variance was predicted.A version of this paper was presented at the Southern Society for Philosophy and Psychology, March 1978.     

Fodor on concepts and Frege Puzzles

Fodor characterizes concepts as consisting of two dimensions: one is content, which is purely denotational/broad, the other the Mentalese vehicle bearing that content, which Fodor calls the mode of presentation (MOP), understood "syntactically." I argue that, so understood, concepts are not interpersonally shareable; so Fodor's own account violates what he calls the Publicity Constraint in his (1998) book. Furthermore, I argue that Fodor's non-semantic solution to Frege cases succumbs to the problem of providing interpersonally applicable functional roles for MOPs. This is a serious problem because Fodor himself has argued extensively that if Fregean senses or meanings are understood as functional/conceptual roles, then they can't be public, since, according to Fodor, there are no interpersonally applicable functional roles.     

Why are we good at detecting cheaters? A reply to Fodor

It is argued that the recent criticism by Fodor (Cognition 75 (2000) 29) of "cheater detection" in the Wason selection task is based upon a false presumption about what the task entails. Fodor compares two different ways of presenting the task, rather than two different task domains (social and non-social). Consequently, the conclusion that the selection task can tell us nothing about either the architecture or the history of cognition is invalid. Fodor's explanation of the Wason selection task is examined experimentally and compared to predictions derived from social contract theory (Cognition 31 (1989) 187). It is concluded that, although Fodor's variant of the Wason selection task improves performance, this improvement is independent of the task domain and is insufficient to account for the "cheater detection" effect.     

Phenomenal Concepts and Higher-Order Experiences

Relying on a range of now-familiar thought-experiments, it has seemed to many philosophers that phenomenal consciousness is beyond the scope of reductive explanation. (Phenomenal consciousness is a form of state-consciousness, which contrasts with creature-consciousness, or perceptual -consciousness. The different forms of state-consciousness include various kinds of access-consciousness, both first-order and higher-order–see Rosenthal, 1986; Block, 1995; Lycan, 1996; Carruthers, 2000. Phenomenal consciousness is the property that mental states have when it is like something to possess them, or when they have subjectively-accessible feels; or as some would say, when they have qualia (see fn. l below).) Others have thought that we can undermine the credibility of those thought-experiments by allowing that we possess purely recognitional concepts for the properties of our conscious mental states. This paper is concerned to explain, and then to meet, the challenge of showing how purely recognitional concepts are possible if there are no such things as qualia –in the strong sense of intrinsic (non relational, non-intentional) properties of experience. It argues that an appeal to higher-order experiences is necessary to meet this challenge, and then deploys a novel form of higher-order thought theory to explain how such experiences are generated.     

Structures of greater good theodicies: The objection from alternative goods

Conclusion The main issue in this paper has been whether appeal to greater goods can explain why God, if he exists, is justified in refraining from ensuring that there is little or no evil. I have argued that in principle it can. I have not defended the plausibility of the view that there actually are greater goods for whose sake God’s actions are planned; nor have I attempted to identify candidate goods performing the role envisaged. However those tasks form part of my larger project. Well beyond my reach is an item on my idle wish list: an explanation of why, assuming that God could have strongly actualized suitable alternative goods with less evil, he did not in fact do so. Address delivered at the Society for Philosophy of Religion and Philosophical Theology (publishers ofSophia), annual meeting, February 1998, in the University of Melbourne.     

Reply to Louise Antony

In her discussion of Naomi Scheman's “Individualism and the Objects of Psychology” Louise Antony misses the import of an unpublished paper of Scheman's that she cites. That paper argues against token identity theories on the grounds that only the sort of psycho-physical parallelisms that token identity theorists, such as Davidson and Fodor, reject could license the claim that each mental state or event is some particular physical state or event.     

Directions in Connectionist Research: Tractable Computations Without Syntactically Structured Representations

Should connectionists abandon the quest for tractably computable cognitive transition functions while retaining syntactically structured mental representations? We argue, in opposition to Horgan, that it should not. We argue that the case against tractably computable functions, based upon the claimed isotropic and Quinean character of cognition, fails since cognition is not as isotropic and Quinean as Fodor and Horgan contend. Moreover, we illustrate how current research in both connectionism and cognitive neuroscience suggests that tractability can be preserved through division of overall computations into modular sub-processes, each of which is tractable. As to syntactically structured representations, we argue that they are unneeded for most cognitive tasks organisms confront, and that when they are needed, they may be provided by external representational media such as natural language. Moreover, we note that increasingly cognitive linguistics has become the ally of connectionism and that the research program of cognitive linguistics suggests that abilities to use natural languages may be developed without requiring syntactically structured mental representations to exist prior to natural language.