Jon Waskan

Figure 1: A pr ototyp ical exa mple of a three-layer feed forward network, used by Plunkett and M archm an (1 991 ) to simulate learning the past-tense of En glish verbs. The inpu t units encode representations of the three phonemes of the present tense of the artificial words used in this simulation. Th e netwo rk is trained to produce a representation of the phonemes employed in the past tense form and the suffix (/d/, /ed/, or /t/) used on regular verbs. To run the network, each input unit is assigned an activation value o f 0 or 1 , dep ending on whethe r the featu re is present or not. Eac h input unit is conne cted to each of the 30 hidden units by a we ighted conn ection and p rovid es an inp ut to each hidden unit equal to the product of the input unit's activation and the weight. Each hidd en unit's activation is then determined by summing ov er the va lues co ming fro m each inp ut unit to deter mine a netinput, and then applying a non-linear function (e.g., the logistic function 1/(1+e^netinput). Th is whole proced ure is.

Jonathan Walkan challenges cognitive science's dominant model of mental representation and proposes a novel, well-devised alternative. The traditional view in the cognitive sciences uses a linguistic model of mental representation. That logic-based model of cognition informs and constrains both the classical tradition of artificial intelligence and modeling in the connectionist tradition. It falls short, however, when confronted by the frame problem---the lack of a principled way to determine which features of a representation must be updated when new information becomes available. So far, proposed alternatives, including the imagistic model, have not resolved the problem. Waskan proposes the Intrinsic Cognitive Models hypothesis, according to which representational states can be concpetualized as the cognitive equivalent of scale models.Waskan argues further that the proposal that humans harbor and manipulate cognitive counterparts to scale models offers the only viable explanation for what most clearly differentiates humans from other creatures: the capacity to engage in truth-preserving manipulation of representations. The ICM hypothesis, he claims, can be distinguished from sentence-based accounts of truth preservation in a way that is fully compatible with what is known about the brain

Theories concerning the structure, or format, of mental representation should (1) be formulated in mechanistic, rather than metaphorical terms; (2) do justice to several philosophical intuitions about mental representation; and (3) explain the human capacity to predict the consequences of worldly alterations (i.e., to think before we act). The hypothesis that thinking involves the application of syntax-sensitive inference rules to syntactically structured mental representations has been said to satisfy all three conditions. An alternative hypothesis is that thinking requires the construction and manipulation of the cognitive equivalent of scale models. A reading of this hypothesis is provided that satisfies condition (1) and which, even though it may not fully satisfy condition (2), turns out (in light of the frame problem) to be the only known way to satisfy condition (3).

Much of the philosophical discussion of explanations has centered around two broad conceptions of what sorts of ‘things’ explanations are – namely, the descriptive and ontic conceptions. Defenders of each argue that scientific psychology has at best little to contribute to the study of explanations. These anti-psychologistic arguments come in two main varieties, the metaphysical and the epistemic. Both varieties trace back to Hempel and recur in the more recent writings of prominent mechanists. The metaphysical arguments attempt to combat psychologism about explanation by doubly dissociating explanations themselves from the commonly associated phenomenology of explanation. The epistemic arguments attempt to combat psychologism by doubly dissociating the good-making features of explanations from the CAPE. I focus on the metaphysical arguments here, for only they directly concern the nature of, as opposed to the evaluation conditions for, explanations. The metaphysical arguments are all shown to equivocate over ‘explanation’, and thus they are no threat to the proposal that sometimes explanations just are psychological events. Such arguments do, nonetheless, gesture towards something important – namely, that the CAPE is neither necessary nor sufficient for explanation in any sense. Similar considerations lead another latter-day mechanist to conclude that no phenomenology of any sort is essential for explanation. What all of these thinkers overlook, however, is that there is another important phenomenological correlate of explanation – namely, finding a happening intelligible or understanding how or why-possibly – and it may well be of the very essence of explanation, at least in the psychological sense. I propose a psychomechanistic theory of intelligibility that accounts for how we can, through finite means, make boundless, endlessly qualified inferences about the implications of our explanations. In closing I note that, although finding a happening intelligible does sometimes engender an unjustified confidence in correctness, we should not overlook how attaining this state does at least put us in a kind of epistemic base camp. I also hint at how intelligibility could turn out to be the ultimate source of the descriptive and ontic conceptions of explanation.

Many philosophers of science believe that empirical psychology can contribute little to the philosophical investigation of explanations. They take this to be shown by the fact that certain explanations fail to elicit any relevant psychological events (e.g., familiarity, insight, intelligibility, etc.). We report results from a study suggesting that, at least among those with extensive science training, a capacity to render an event intelligible is considered a requirement for explanation. We also investigate for whom explanations must be capable of rendering events intelligible and whether or not accuracy is also viewed as a requirement.

Many philosophers of science follow Hempel in embracing both substantive and methodological anti-psychologism regarding the study of explanation. The former thesis denies that explanations are constituted by psychological events, and the latter denies that psychological research can contribute much to the philosophical investigation of the nature of explanation. Substantive anti-psychologism is commonly defended by citing cases, such as hyper-complex descriptions or vast computer simulations, which are reputedly generally agreed to constitute explanations but which defy human comprehension and, as a result, fail to engender any relevant psychological events. It is commonly held that the truth of the substantive thesis would lend support to the methodological thesis. However, the standard argument for the substantive thesis presumes that philosophers’ own judgments about the aforementioned cases issues from mastery of the lay or scientific norms regarding the use of ‘explanation.’ Here we challenge this presumption with a series of experiments indicating that both lay and scientific populations require of explanations that they actually render their targets intelligible. This research not only undermines a standard line of argument for substantive anti-psychologism, it demonstrates the utility of psychological research methods for answering meta-questions about the norms regarding the use of ‘explanation.’.

Here I consider the relative merits of two recent models of explanation, James Woodward's interventionist-counterfactual model and the model model. According to the former, explanations are largely constituted by information about the consequences of counterfactual interventions. Problems arise for this approach because countless relevant interventions are possible in most cases and because it overlooks other kinds of equally relevant information. According the model model, explanations are largely constituted by cognitive models of actual mechanisms. On this approach, explanations tend not to represent any of the aforementioned information explicitly but can instead be used to produce it on demand. The model model thus offers the more plausible account of the information of which we are aware when we have an explanation and of the ratiocinative process through which we derive many kinds of information that are relevant to the evaluation of explanations.

Summary. The viability of the proposal that human cognition involves the utilization of nonsentential models is seriously undercut by the fact that no one has yet given a satisfactory account of how neurophysiological circuitry might realize representations of the right sort. Such an account is offered up here, the general idea behind which is that high-level models can be realized by lower—level computations and, in turn, by neural machinations. It is shown that this account can be usefully applied to deal with problems in fields ranging from artificial intelligence to the philosophy of science

Resurgent interest in both mechanistic and counterfactual theories of explanation has led to a fair amount of discussion regarding the relative merits of these two approaches. James Woodward is currently the pre-eminent counterfactual theorist, and he criticizes the mechanists on the following grounds: Unless mechanists about explanation invoke counterfactuals, they cannot make sense of claims about causal interactions between mechanism parts or of causal explanations put forward absent knowledge of productive mechanisms. He claims that these shortfalls can be offset if mechanists will just borrow key tenets of his counterfactual theory of causal claims. What mechanists must bear in mind, however, is that by pursuing this course they risk both the assimilation of the mechanistic theories of explanation into Woodward’s own favored counterfactual theory, and they risk the marginalization of mechanistic explanations to a proper subset of all explanations. An outcome more favorable to mechanists might be had by pursuing an actualist-mechanist theory of the contents of causal claims. While it may not seem obvious at first blush that such an approach is workable, even in principle, recent empirical research into causal perception, causal belief, and mechanical reasoning provides some grounds for optimism.

Cognitive Science is a major new guide to the central theories and problems in the study of the mind and brain. The authors clearly explain how and why cognitive science aims to understand the brain as a computational system that manipulates representations. They identify the roots of cognitive science in Descartes - who argued that all knowledge of the external world is filtered through some sort of representation - and examine the present-day role of Artificial Intelligence, computing, psychology, linguistics and neuroscience. Throughout, the key building blocks of cognitive science are clearly illustrated: perception, memory, attention, emotion, language, control of movement, learning, understanding and other important mental phenomena. Cognitive Science: presents a clear, collaborative introduction to the subject is the first textbook to bring together all the different strands of this new science in a unified approach includes illustrations and exercises to aid the student

I have argued elsewhere that non-sentential representations that are the close kin of scale models can be, and often are, realized by computational processes. I will attempt here to weaken any resistance to this claim that happens to issue from those who favor an across-the-board computational theory of cognitive activity. I will argue that embracing the idea that certain computers harbor nonsentential models gives proponents of the computational theory of cognition the means to resolve the conspicuous disconnect between the sentential character of the data structures they posit and the nonsentential qualitative character of our perceptual experiences of corporeal (i.e., spatial, kinematic, and dynamic) properties. Along the way, I will question the viability of some externalist remedies for this disconnect, and I will explain why the computational theory put forward here falls quite clearly beyond the useful bounds of the Chinese-Room argument.