The fields of machine learning (ML) and cognitive science have developed complementary approaches to computationally modeling human behavior. ML's primary concern is maximizing prediction accuracy; cognitive science's primary concern is explaining the underlying mechanisms. Cross-talk between these disciplines is limited, likely because the tasks and goals usually differ. The domain of e-learning and knowledge acquisition constitutes a fruitful intersection for the two fields’ methodologies to be integrated because accurately tracking learning and forgetting over time and predicting future performance based on (...) learning histories are central to developing effective, personalized learning tools. Here, we show how a state-of-the-art ML model can be enhanced by incorporating insights from a cognitive model of human memory. This was done by exploiting the predictive performance equation's (PPE) narrow but highly specialized domain knowledge with regard to the temporal dynamics of learning and forgetting. Specifically, the PPE was used to engineer timing-related input features for a gradient-boosted decision trees (GBDT) model. The resulting PPE-enhanced GBDT outperformed the default GBDT, especially under conditions in which limited data were available for training. Results suggest that integrating cognitive and ML models could be particularly productive if the available data are too high-dimensional to be explained by a cognitive model but not sufficiently large to effectively train a modern ML algorithm. Here, the cognitive model's insights pertaining to only one aspect of the data were enough to jump-start the ML model's ability to make predictions—a finding that holds promise for future explorations. (shrink)

The fields of machine learning (ML) and cognitive science have developed complementary approaches to computationally modeling human behavior. ML's primary concern is maximizing prediction accuracy; cognitive science's primary concern is explaining the underlying mechanisms. Cross-talk between these disciplines is limited, likely because the tasks and goals usually differ. The domain of e-learning and knowledge acquisition constitutes a fruitful intersection for the two fields’ methodologies to be integrated because accurately tracking learning and forgetting over time and predicting future performance based on (...) learning histories are central to developing effective, personalized learning tools. Here, we show how a state-of-the-art ML model can be enhanced by incorporating insights from a cognitive model of human memory. This was done by exploiting the predictive performance equation's (PPE) narrow but highly specialized domain knowledge with regard to the temporal dynamics of learning and forgetting. Specifically, the PPE was used to engineer timing-related input features for a gradient-boosted decision trees (GBDT) model. The resulting PPE-enhanced GBDT outperformed the default GBDT, especially under conditions in which limited data were available for training. Results suggest that integrating cognitive and ML models could be particularly productive if the available data are too high-dimensional to be explained by a cognitive model but not sufficiently large to effectively train a modern ML algorithm. Here, the cognitive model's insights pertaining to only one aspect of the data were enough to jump-start the ML model's ability to make predictions—a finding that holds promise for future explorations. (shrink)

The parameters governing our behavior are in constant flux, and capturing these dynamics in cognitive models remains a challenge. We demonstrate how a mapping between ACT‐R's model of declarative memory and the linear ballistic accumulator enables efficient estimation of memory parameters from data. The resulting estimates provide a cognitively meaningful explanation for observed differences in behavior over time and between individuals.

The parameters governing our behavior are in constant flux. Accurately capturing these dynamics in cognitive models poses a challenge to modelers. Here, we demonstrate a mapping of ACT-R's declarative memory onto the linear ballistic accumulator (LBA), a mathematical model describing a competition between evidence accumulation processes. We show that this mapping provides a method for inferring individual ACT-R parameters without requiring the modeler to build and fit an entire ACT-R model. Existing parameter estimation methods for the LBA can be used, (...) instead of the computationally expensive parameter sweeps that are traditionally done. We conduct a parameter recovery study to confirm that the LBA can recover ACT-R parameters from simulated data. Then, as a proof of concept, we use the LBA to estimate ACT-R parameters from an empirical dataset. The resulting parameter estimates provide a cognitively meaningful explanation for observed differences in behavior over time and between individuals. In addition, we find that the mapping between ACT-R and LBA lends a more concrete interpretation to ACT-R's latency factor parameter, namely as a measure of response caution. This work contributes to a growing movement towards integrating formal modeling approaches in cognitive science. (shrink)

We face reality presented with the data of conscious experience and nothing else. The project of early modern philosophy was to build a complete theory of the world from this starting point, with no cheating. Crucial to this starting point is the data of conscious sensory experience – sense data. Attempts to avoid this project often argue that the very idea of sense data is confused. But the sense-data way of talking, the (...) sense-data language, can be freed from every blemish using ideas from contemporary metaontology. We can adopt a sense-data framework that vindicates the traditional claims of sense-data theories and leads to plausible theories of perception and color. We can, we should, and in a sense, we must. Yet when we do, we face the traditional problem of external world skepticism, head-on. The real challenge of skepticism is forced upon us; it cannot honestly be avoided using externalist tricks, burden of proof shifting, or other razzle dazzle. But the challenge can be met: we are rational to posit the external world as the best explanation of the many synchronic and diachronic patterns over our sense-data. This paper argues for all of these points and ends with a plea for analytic empiricism – a traditional sense-data version of empiricism that uses all of the tools of analytic philosophy while avoiding the more questionable doctrines of the Vienna Circle (phenomenalism, verificationism). Analytic empiricism is our last best hope for completing the great philosophical project of early modern philosophy. (shrink)

In this paper, I present a version of a sense-data approach to perception, which differs to a certain extent from well-known versions like the one put forward by Jackson. I compare the sense-data view to the currently most popular alternative theories of perception, the so-called Theory of Appearing (a very specific form of disjunctivist approaches) on the one hand and reductive representationalist approaches on the other. I defend the sense-data approach on the basis that (...) it improves substantially on those alternative theories. (shrink)

The theory of knowledge in early twentieth-century Anglo American philosophy was oriented toward phenomenally described cognition. There was a healthy respect for the mind-body problem, which meant that phenomena in both the mental and physical domains were taken seriously. Bertrand Russell's developing position on sense-data and momentary particulars drew upon, and ultimately became like, the neutral monism of Ernst Mach and William James. Due to a more recent behaviorist and physicalist inspired "fear of the mental", this development has (...) been down-played in historical work on early analytic philosophy. Such neglect assumes that the "linguistic turn" is a proper and permanent effect of twentieth-century philosophy, an assumption that distorts early analytic historiography, and begs a substantive philosophical question about thought and cognition. (shrink)

Sense data are the alleged mind-dependent objects that we are directly aware of in perception, and that have exactly the properties they appear to have. For instance, sense data theorists say that, upon viewing a tomato in normal conditions, one forms an image of the tomato in one's mind. This image is red and round. The mental image is an example of a “sense datum.” Many philosophers have rejected the notion of sense data, (...) either because they believe that perception gives us direct awareness of physical phenomena, rather than mere mental images, or because they believe that the mental phenomena involved in perception do not have the properties that appear to us (for instance, I might have a visual experience representing a red, round tomato, but my experience is not itself red or round). Defenders of sense data have argued, among other things, that sense data are required to explain such phenomena as perspectival variation, illusion, and hallucination. Critics of sense data have objected to the theory's commitment to mind-body dualism, the problems it raises for our knowledge of the external world, its difficulty in locating sense data in physical space, and its apparent commitment to the existence of objects with indeterminate properties. (shrink)

The first two sections of the paper characterize the nineteenth century respect for the phenomenal by considering Helmholtz’s position and James’ and Russell’s move to neutral monism. The third section displays a moment’s sympathy with those who recoiled from the latter view -- but only a moment’s. The recoil overshot what was a reasonable response, and denied the reality of the phenomenal, largely in the name of the physical or the material. The final two sections of the paper develop a (...) third way, which retains a healthy respect for the mental and for the mind–body relation, does not attempt to equate objects with congeries of sensations, and does not attempt to deny the reality of the phenomenal. In fact, I will claim that on some conceptions (and not merely idealist-phenomenalist conceptions), the phenomenal is a fact of nature, and hence a part of the natural world. Some aspects of this third way are familiar in the various representational and critical realisms of the twentieth-century. But the realization -- or, more neutrally, the conception -- that the natural might include the phenomenal is less familiar. Yet this position has its predecessors too, not only among the physicists and psychologists of the nineteenth century, but among major physicists (as opposed to physicalist philosophers) and psychologists of the twentieth. [A re-edited version of this paper appears in Gary Hatfield, Perception and Cognition: Essays in the Philosophy of Psychology, Clarendon Press, 2009.]. (shrink)

Though scholarship has explored Karin Costelloe-Stephen’s contributions to the history of psychoanalysis, as well as her relations to the Bloomsbury Group, her philosophical work has been almost completely ignored. This paper will examine her debate with Bertrand Russell over his criticism of Bergson. Costelloe-Stephen had employed the terminology of early analytic philosophy in presenting a number of arguments in defence of Bergson’s views. Costelloe-Stephen would object, among other things, to Russell’s use of an experiment which, as she points out, was (...) first conducted by Carl Stumpf. Russell appeals to Stumpf's experiment in his attempt to prove that sense data are terms in logical relations, a thesis presupposed by the project of logical analysis outlined in Our Knowledge of the External World. A reformulated version of Costelloe-Stephen's argument put forth by this paper shows that Russell's argument fails to provide adequate proof for his thesis. Further modifications of the argument can also address a reconstruction (based on contemporary reports) of Russell's reply to Costelloe-Stephen. In his reply, Russell would use, already in 1914, the term ‘analytic philosophy’ in contrasting his and Moore’s approach to a continental one, exemplified by Bergson and Costelloe-Stephen. (shrink)

Additional arguments for sense‐data begin by defending the claim that perceptual sensations are psychological individuals, examples being phosphenes, after‐images, and the ‘ringings’ of ‘tinnitus’. Five arguments for sense‐data follow. First, that since corresponding to every veridical visual field is a possible non‐veridical visual field of sensations, the latter merely needs a different and regular outer cause to be deemed veridical. Second, since bodily sensation experience is extremely strong evidence for the existence of a matching sensation cause, (...) the experience of ‘ringing’ must be strong evidence that a ‘ringing’ sensation is its cause, and this holds in the veridical hearing of a ringing sound. Third, that it is inconceivable that one has a truly visual experience of a red visual field, and absolutely nothing red‐looking be there before one had noticed. Fourth, that the experience of sound might have had a different developmental history, in which first there was merely experience of auditory sensation, and then over millennia such experience gradually found itself caused by regular outer phenomena, thereby constituting the hearing of sound. And a fifth argument that resembles the first argument. (shrink)

When I began to study philosophy sense-data were in the fashion; everybody had some. Nowadays talking about sense-data, like distinguishing between “shall” and “will”, is apt to be regarded as an indication that one has stopped moving with the times. Before abandoning this old habit, I want to consider whether there may not after all be something in a doctrine adopted by so many leading philosophers in pre-war England.

The following remarks have been stimulated by Mr. A. J. Ayer's recent essay “The Terminology of Sense Data.” In this paper Mr. Ayer restates several of the points he has made in his book The Foundations of Empirical Knowledge. The context of his argument is that of the traditional distinction between two kinds of things, sense data, percepts, phenomenal or direct given-nesses on the one hand and so-called physical objects on the other. In this context, Mr. (...) Ayer's argument is, to my mind, very admirable; so I do not intend to comment on it. In some other contexts, some of his formulations may, I fear, give rise to misunderstandings; these, naturally, are the points on which I wish to comment. But before specifying them, I had better make sure that we are talking about the same thing by briefly restating several of Mr. Ayer's points. With some of these, if I understand them correctly, I find myself in agreement; the merit of some others I shall presently examine. (shrink)

Philosophers have given various reasons for denying the existence of sense data. A number of these reasons are examined in the present paper. The claim that ?no sufficient purpose is served by positing such objects? is deemed irrelevant to the issue; the complaint that ?we do not know what it would be like to find that there were no such objects? is found to be confusedly formulated, mistaken, and irrelevant; and the charge that there is something improper, extraordinary, (...) or defective about the way in which philosophers have introduced the term ?sense datum? into their discourse is pronounced false. (shrink)

This paper examines and defends the view that the immediate objects of visual perception, or what are often called sense data, are parts of the facing surfaces of physical objects-the naturalized sense data (NSD) theory. Occasionally defended in the literature on the philosophy of perception, most famously by G. E. Moore (1918-1919), it has not proved popular and indeed was abandoned by Moore himself. The contemporary situation in the philosophy of perception seems ripe for a revaluation (...) of the NSD theory. however. The NSD theory allows us to accommodate the very real shortcomings in uncritical direct realism without postulating the existence of non-physical sense data in a way that has seemed to many incompatible with any robust form of philosophical naturalism.The argument to establish the NSD theory proceeds in two stages. In §II I argue against the direct realist that we perceive three-dimensional material objects in virtue of perceiving parts of their surfaces. The argument for this conclusion involves clearly distinguishing (in § I) between two notions that have tended to be run together in discussions of perception---namely, immediate perception and direct perception. In §III I argue against the sense-datum theorist that those parts of the surfaces of those objects are not themselves perceived in virtue of the perception of anything else. (shrink)

Experiences of all kinds have a distinctive character, which marks them out as intrinsically different from states of consciousness such as thinking. A plausible view is that the difference should be accounted for by the fact that, in having an experience, the subject is somehow immediately aware of a range of phenomenal qualities. For example, in seeing, grasping and tasting an apple, the subject may be aware of a red and green spherical shape, a certain feeling of smoothness to touch, (...) and a sweet sensation. Such phenomenal qualities are also immediately present in hallucinations. According to the sense-data theory, phenomenal qualities belong to items called “sense-data.” In having a perceptual experience the subject is directly aware of, or acquainted with, a sense-datum, even if the experience is illusory or hallucinatory. The sense-datum is an object immediately present in experience. It has the qualities it appears to have. (shrink)

Although it has been something of a fetish for philosophers to distinguish between hallucination and illusion, the enduring problems for philosophy of perception that both phenomena present are not essentially different. Hallucination, in its pure philosophical form, is just another example of the philosopher’s penchant for considering extreme and extremely idealized cases in order to understand the ordinary. The problem that has driven much philosophical thinking about perception is the problem of how to reconcile our evident direct perceptual contact with (...) objects and properties with the equally evident fact that there is no phenomenological signal separating error and truth. “The obscure object of hallucination” offers a subtle and plausible solution to this problem and one that solves the problem generally, not just in the special case of hallucination. Johnston’s objective is to offer a theory of perception that meets two constraints: (1) that it provide an explanation of the possibility of delusive and veridical sensings that are indistinguishable from the first-person perspective and (2) that it count as form of direct realism where this is taken to involve acquaintance with the objects of perception. Johnston uses the first constraint to rule out disjunctivism. The second constraint is used to rule out conjunctivism, which as Johnston uses the term, includes most of the widely adopted philosophical theories of perception. Johnston also develops his own sophisticated and interesting theory of perception. In what follows, I will discuss the relation of Johnston’s theory to conjunctivism, examine one of his anti-conjunctivist arguments and finally compare Johnston’s theory with some other versions of direct realism. These topics constitute a very incomplete selection of the important issues discussed in this rich and interesting paper. I will also not disagree, in any fundamental way, with any of the central theses of Johnston’s discussion.. (shrink)

Analytic philosophy arose in the early decades of the twentieth century, with Bertrand Russell and G. E. Moore leading the way. Although some accounts emphasize the role of logic and language in the origin of analytic philosophy, of equal importance is the theme of perception, sense data, and knowledge, which dominated systematic philosophical discussion in the first two decades of the twentieth century in both Britain and America. This chapter examines work on perception and sense data (...) as well as the external world in the analytic tradition in the first half of the twentieth century. After sketching the situation before and just after 1900, it turns to Moore, Russell, and their interlocutors in the teens and twenties, addresses the American scene, looks to subsequent developments in the 1930s and beyond, and finally considers the fates of sense data and the ‘given’ and of the topic of perception more generally. (shrink)

This paper concerns the structure of appearances. I argue that to be appeared to in a certain way is to be aware of one or more universals. Universals therefore function like the sense‐data, once highly favoured but now out of fashion. For instance, to be appeared to treely, in a visual way, is to be aware of the complex relation, being tree‐shaped and tree‐coloured and being in front of, a relation of a kind which could be instantiated by (...) a material object and a perceiver, which is thus instantiated in the veridical case but not in the non‐veridical. (shrink)

It is important for our purposes to notice that in this first reduction of Theætetus' definition of knowledge as perception, Plato has introduced the distinction between sense object and physical object, for he has specifically said, "when the same wind is blowing, one of us feels chilly, the other does not." In using this example. Plato has, as Cornford observes, raised the question of how the several sense objects are related to the single physical object. This question is (...) one of the two major questions with which this paper will deal, since it is, as I have remarked, the more important question concerning the ontological status of sense-data. Cornford feels that Protagoras probably held that the physical object is both hot and cold, different observers perceiving different aspects of the same object. That is, Cornford suggests that Protagoras' own answer to this question of the relation between sense-data and physical objects is one of complete identification in every case. We will see that this theory is similar in many ways to the one worked out by Plato in the Theætetus, but the distinction between the physical object and the sense object, between the wind which feels hot and cold to different people, makes certain characteristic differences between Plato's and Protagoras' theory. Before going on to examine these differences, it is necessary to follow the further development of Plato's general theory of perception. (shrink)

Is the sense‐data theory, otherwise known as indirect realism, a form of representationalism? This question has been under‐explored in the extant literature, and to the extent that there is discussion, contemporary authors disagree. There are many different variants of representationalism, and differences between these variants that some people have taken to be inconsequential turn out to be key factors in whether the sense‐data theory is a form of representationalism. Chief among these are whether a representationalist takes (...) the phenomenal character of an experience to be explicable in virtue of the properties of an experience that represent something or explicable in virtue of that which gets represented. Another is whether representationalists hold a non‐reductionist, or naturalistically or non‐naturalistically reductionist variant of representationalism. In addition, subtle differences in what one takes phenomenal character to be on the sense‐data theory – either awareness of sense‐data or the sense‐data themselves – together with one's account of representation, are crucial factors in determining whether sense‐data theory is compatible with representationalism. This paper explores these relationships and makes manifest the complexities of the metaphysics of two central theories of perception. 1 Many thanks are due to Clare Batty, Derek Brown, and James Stazicker for discussion and feedback on earlier drafts, which made this essay better than it would otherwise have been. This work was supported by two grants from the Arts and Humanities Research Council (grant numbers AH/1027509/1 and AH/L007053/1). (shrink)