COGNITIVE SCIENCE

Below are readings and essay questions for tutorials in Philosophy of Cognitive Science. Many of the readings are online, and all can be obtained from the college or other libraries in Oxford, but if you are struggling to get hold of anything, email me, as I have PDF copies of nearly everything.

The default plan is to cover the following TUTORIAL TOPICS. We begin by looking at the metatheory of cognitive science, focusing on classic work by Chomsky in linguistics and Marr on vision. These first two weeks provide some theoretical background to the next two topics, the classical computational theory of mind, as it is articulated by one of its leading proponents, Jerry Fodor, and one of the main alternatives to it, connectionism. We then look at modularity and consciousness, exploring the potential limits of cognitive science. The topic of the last week is left open, giving you scope to pursue your own areas of interest. Options here include, but are not limited to, the OTHER TOPICS.

The reading for each topic is divided into two parts. In writing your tutorial essay, focus on the CORE READING, using as a guide the more introductory texts marked with a star (*) and any Faculty lectures on the topic—the lectures may be available online via Canvas, and are a good indication of what might come up in the exams. You can then look at the FURTHER READING, as well as anything else on the Faculty Reading List, when exploring topics in more depth during later vacations.

The current version of this reading list was put together in light of my experience using previous incarnations in teaching Philosophy of Cognitive Science to undergraduates in Oxford over the years. If you’d like to use the list for teaching, please feel free. Feedback and corrections are gratefully received!

TUTORIAL TOPICS

1. Knowledge of Language
2. Marr’s Levels of Analysis
3. The Language of Thought Hypothesis
4. Connectionism: Systematicity
5. Modularity I: Input Systems
6. Modularity II: Central Systems
7. Consciousness: Access and Phenomenal
8. TBA

OTHER TOPICS

• Cognitive Neuropsychology
• Neuroimaging
• Delusions
• Understanding Minds
• Evolutionary Psychology
• Free Will
• Concepts
• Innateness
• The Frame Problem
• Psychological Explanation

ANTHOLOGIES and TEXTBOOKS

The following anthologies are particularly useful, and contain many of the key readings. I refer to them below as B, C&C, and H respectively:

There is no set textbook, but the following are all recommended:

Martin Davies (2009) ‘Cognitive Science’, in Frank Jackson and Michael Smith, eds. The Oxford Handbook of Contemporary Philosophy (OUP), surveys much of what we’ll cover, and is recommended as reading in the vacation beforehand. An expanded version is on WebLearn.

TUTORIAL TOPICS

1. KNOWLEDGE of LANGUAGE

TOP

ESSAY QUESTION

Is knowledge of language constituted by tacit knowledge of linguistic rules?

CORE READING

FURTHER READING

Chomsky’s work in linguistics was a major part of the cognitive revolution of the 1950s. Chomsky (1965) presents some of his main conceptual innovations, including the notion of tacit knowledge. George (1986) is a short, and extremely useful discussion of the Quine-Chomsky debate. Evans’ dispositional approach was developed by Martin Davies and Christopher Peacocke, and criticised by Crispin Wright. For references and discussion, see their contributions to George (1989). (The contributions from George and Putnam are also relevant to this week’s topic.) Miller (1997) is a useful review of the debate, and there is some critical discussion of the dispositional approach in Collins (2004), who argues Chomsky’s notion of tacit knowledge is not epistemological in any substantive sense. Collins is responding to Jerry Fodor’s account of how Chomsky’s notion of a language faculty differs from Fodor’s own notion of a language module. See Fodor (1983)—but we’ll look more closely at Fodor’s views on modularity later.

PAST PAPER QUESTIONS

Does knowing how to speak a language consist in tacit knowledge of linguistic rules? (2020)

What role, if any, does tacit knowledge play in language acquisition? (2019)

What, if anything, is the difference between our knowledge of language and our knowledge of a memorized look-up table (like a phone book)? (2018)

Does our knowledge of language differ fundamentally from our knowledge of a list of arbitrary facts, such as the winning lottery numbers in the UK for the past ten years? (2017)

2. MARR’S LEVELS of ANALYSIS

TOP

ESSAY QUESTION

What are Marr’s three levels of analysis, and why did he think we need them? Which of them are most important for cognitive psychology, and in what sense, if any, are they independent of one another?

CORE READING

FURTHER READING

Peacocke (1986) argues for the need for a fourth level, intermediate between Marr’s computational and algorithmic-representational levels. See also the commentaries on this, and Peacocke’s replies, in Mind & Language Volume 1, Issue 4. Sterelny (1990) is introductory, discussing problems in determining the algorithmic-representational level and the distinction between psychology and neuroscience. For more on the latter issue, see Gold and Stoljar (1999), published alongside various peer commentaries, and Feest (2003). Kitcher (1988), though more focused on Marr’s theory of vision, is also relevant. Bechtel and Shagrir (2015), in the CORE READING, is part of a special issue of Topics in Cognitive Science marking the 30th anniversary of Marr’s Vision, and many of the other papers in the issue are useful too.

PAST PAPER QUESTIONS

‘Marr’s three levels show that neuroscience cannot be done in isolation from computational theory.’ Discuss. (2020)

EITHER
(a) What are Marr’s three levels? Can we fruitfully study the mind in any one of these levels independently of the other two?
OR
(b) What would it be to explain a cognitive capacity in terms of functional decomposition? How would such explanations differ from explanations in terms of laws? (2019)

EITHER
(a) ‘Marr’s three levels are separate, therefore it is of no use to study one level if one wishes to say something about another level.’ Discuss.
OR
(b) Does any one of Marr’s levels have priority with respect to providing an explanation for a cognitive system overall and with respect to guiding research at all (other) levels? (2018)

EITHER
(a) How do Marr’s levels relate to each other?
OR
(b) ‘The personal/subpersonal distinction is best understood as a distinction between two types of psychological explanation – “horizontal” and “vertical”.’ (DRAYSON) Explain and discuss. (2017)

3. THE LANGUAGE of THOUGHT HYPOTHESIS

TOP

ESSAY QUESTION

What is the Language of Thought Hypothesis (LOTH)? How does it differ from the Representational and Computational Theories of Mind (RTM and CTM)? What are the best arguments for and against it?

CORE READING

FURTHER READING

Fodor (1975) sets out and defends the LOTH alongside the RTM and CTM. Dennett (1976) is an influential review, criticising Fodor’s commitment to the RTM. Rey (1995) offers a more or less a priori line of argument for the LOTH. Knowles (1998) is a critical discussion of Laurence and Margolis (1997), who reply in Margolis and Laurence (1999). For a retrospective from Fodor, responding to regress objections and defending his thesis that some concepts are innate, see his (2008). For a defence of the map-like view discussed in Braddon-Mitchell and Jackson (2007), see Camp (2007). For more on the productivity and systematicity arguments for the LOTH, see next week’s topic, CONNECTIONISM.

PAST PAPER QUESTIONS

EITHER
(a) Do we need to posit a Language of Thought?
OR
(b) Can connectionists explain the compositionality of thought? (2020)

Evaluate the argument from systematicity for the Language of Thought Hypothesis. (2019)

Do the productivity and systematicity of thought provide arguments for a certain kind of view on which mental representations have structure? (2018)

What is the language of thought hypothesis and does it entail that thought has all the properties of a language like English? (2017)

What commitments does the language of thought hypothesis have beyond a basic commitment to understanding cognition as information processing? (2016)

4. CONNECTIONISM: SYSTEMATICITY

TOP

ESSAY QUESTION

‘Connectionist models of cognitive processes are either unable to account for the systematicity of cognitive processes or are merely implementations of classical architectures. Either way, it follows that the mind cannot be a connectionist network.’ Explain and assess this objection.

CORE READING

These are all reprinted in Cynthia Macdonald and Graham Macdonald, eds. Connectionism: Debates on Psychological Explanation (Blackwell), Part I. Alongside them, you will also find a reply to Fodor and McLaughlin (1990) by Paul Smolensky, and a very useful introduction to the debate by Cynthia Macdonald. Part II of the collection contains classic papers addressing the question whether connectionism undermines folk psychology, another issue to explore.

FURTHER READING

Matthews (1997) argues that neither connectionists nor classicists have an explanation of systematicity, but that there are grounds for optimism about the prospects for a connectionist account. Johnson (2004) argues that thought is not systematic, properly understood, so there is nothing here for the connectionist to explain anyway. McLaughlin (2009) responds to both Johnson and Robert Cummins, from whom Johnson took his cue. Calvo and Symons (2014) is a recent volume on the systematicity debate. Garson (1997/2015) is a survey, summarising the systematicity debate and many other issues.

PAST PAPER QUESTIONS

EITHER
(a) Do we need to posit a Language of Thought?
OR
(b) Can connectionists explain the compositionality of thought? (2020)

What, if any, are the advantages of connectionism over the classical computational- representational theory of mind? (2019)

EITHER
(a) What advantages, if any, do symbolic accounts of mental representation have in explaining productivity and systematicity?
OR
(b) ‘I should prefer to explain the sense in which thoughts are structured, not in terms of their being composed of several distinct elements, but in terms of their being a complex of the exercise of several distinct conceptual abilities.’ (EVANS) Discuss. (2017)

What empirical content is there to the claim that thought is productive and systematic? Would the truth of the claim teach us anything about cognitive architecture? (2016)

5. MODULARITY I: INPUT SYSTEMS

TOP

ESSAY QUESTION

Explain and assess Fodor’s view that input systems are modular.

CORE READING

FURTHER READING

If you’re working on this topic (or the next) in more depth, Fodor (1983) is a must-read. You’ll also find lots of useful discussion in the critical responses to the précis, his (1985) piece in the Core Reading. Prinz (2006) is an accessible discussion, offering a range of arguments against modularity. For more on the question of encapsulation/impenetrability, Pylyshyn (1999) is a must-read, while Machery (2015) and Firestone and Scholl (2016) are highlights of the recent literature. See also Shea (2014) for a useful discussion of the distinction between top-down and bottom-up effects.

PAST PAPER QUESTIONS

Are input systems modular? (2020)

Is the mind modular? (2018)

EITHER
(a) In what sense, if any, is the mind modular?
OR
(b) With reference to examples of perceptual processing or language processing (or both), evaluate the claim that input systems are informationally encapsulated. (2017)

What does it mean to say that the mind is modular? What evidence is there to support the claim that it is? (2016)

6. MODULARITY II: CENTRAL SYSTEMS

TOP

ESSAY QUESTION

Explain and assess Fodor’s view that central systems are (a) not modular, and so are (b) computationally intractable.

CORE READING

FURTHER READING

The massive modularity thesis was first espoused in the early 1990s by evolutionary psychologists such as Lena Cosmides and John Tooby, and Dan Sperber. (See the EVOLUTIONARY PSYCHOLOGY entry below for more details.) However, it received its most thorough development in Peter Carruthers’ book, The Architecture of the Mind. Carruthers (2008) is a précis, and elsewhere in the same issue of Mind & Language, you’ll find commentaries from Fiona Cowie, Edouard Machery, and Robert Wilson, along with replies from Carruthers himself. Barrett and Kurzban (2006) is a useful review of arguments for and against massive modularity. (They are proponents.) Rabaglia et al. (2011) present a challenge for massive modularity based on the so-called positive manifold, the fact that individuals who are high scoring in one cognitive domain tend to be high scoring on other cognitive domains. Fodor (2000) updates his case for thinking that central systems are computationally intractable—Ch. 4 discusses the massive modularity thesis. For critical discussion, see Ludwig and Schneider (2008) and Fuller and Samuels (2014).

PAST PAPER QUESTIONS

What is Fodor’s argument against the modularity of central cognition? Should we accept this argument? (2019)

Is the mind modular? (2018)

EITHER
(a) In what sense, if any, is the mind modular?
OR
(b) With reference to examples of perceptual processing or language processing (or both), evaluate the claim that input systems are informationally encapsulated. (2017)

What does it mean to say that the mind is modular? What evidence is there to support the claim that it is? (2016)

7. CONSCIOUSNESS: ACCESS and PHENOMENAL

TOP

ESSAY QUESTION

Is there a distinction between access consciousness and phenomenal consciousness? If so, how are the two related, and what significance does the distinction have for cognitive science?

CORE READING

FURTHER READING

Nagel (1972) is a classic, and essential background, as is Block (2002). For some recent empirical work, see De Gardelle, Sackur, and Kouider (2009). Block (2011) replies to Cohen and Dennett (2011) and Phillips (2011). Phillips (2016) reviews some recent work on the issue. Schneider and Velmans (2017) is the second edition of an extremely useful collection. See especially the papers in Part I, The Problems of Consciousness.

PAST PAPER QUESTIONS

Does phenomenal consciousness overflow access consciousness? (2020)

Are there good reasons to distinguish between phenomenal consciousness and access consciousness in the scientific study of the mind? (2019)

EITHER
(a) ‘Clearly, phenomenal and access consciousness are often separated.’ Discuss and evaluate this claim.
OR
(b) ‘We can study access consciousness empirically, therefore we can resolve the problem of consciousness.’ Discuss and evaluate this claim. (2018)

‘The best explanation for several perceptual phenomena is that phenomenal consciousness and access consciousness can come apart in both directions.’ Discuss. (2017)

OTHER TOPICS

COGNITIVE NEUROPSYCHOLOGY

TOP

ESSAY QUESTION

How can we make inferences from evidence about the impaired performance of patients following brain injury to theories about the structure of the normal cognitive system? Do double dissociations have a special logical role in these inferences?

CORE READING

FURTHER READING

The use of double dissociations in theorising about cognition goes back to the 19th century, and Paul Broca, who identified a region of the brain employed in language processing, though the term double dissociation was first introduced by Hans-Lukas Tueber (1955). Coltheart (1999) is a short, clear statement of one of the key components of his ultra-cognitivist approach, the idea that the mind is composed of a system of domain-specific modules. Juola and Plunkett (2000) is an influential criticism of double dissociation methodology. McGeer (2007) argues on both empirical and philosophical grounds that the ultra-cognitivist approach is fundamentally misguided. Van Orden et al. (2010) argue that double dissociation of reading modules illustrates general problems with the ultra-cognitivists’ assumption of modularity.

PAST PAPER QUESTIONS

EITHER
(a) ‘Distinguishing two modules is the only way to explain cases of double dissociation.’ Discuss.
OR
(b) Can brain imaging studies teach us anything about cognition? (2020)

Do double dissociations provide a privileged kind of evidence for theories of mental architecture? (2019)

How strong is the claim that double dissociation evidence is superior to other kinds of evidence in the study of the mind/brain? (2018)

EITHER
(a) Is evidence from double dissociation categorically superior to other kinds of evidence in the cognitive sciences?
OR
(b) How has the development of implemented computational models of cognitive processes contributed to cognitive psychology and cognitive neuropsychology? (2017)

NEUROIMAGING

TOP

ESSAY QUESTION

How, if at all, can evidence about brain activity be brought to bear on theorising in cognitive psychology?

CORE READING

This might look like a lot of reading for one week, but all the papers are quite short; the longest is only 12 pages.

FURTHER READING

Poldrack (2006) raises some concerns about reverse inference. Logothetis (2008) is a useful overview of fMRI work, written for an interdisciplinary audience. Harman (2010) discusses potential misunderstandings of reports on fMRI research. Machery (2012) compares forward or “function-to-structure” inferences to the inferences based on dissociations in cognitive neuropsychology, and argues that the former are in poorer epistemic standing, while his (2014) paper defends reverse inference.

PAST PAPER QUESTIONS

EITHER
(a) ‘Distinguishing two modules is the only way to explain cases of double dissociation.’ Discuss.
OR
(b) Can brain imaging studies teach us anything about cognition? (2020)

‘Cognitive theories, by definition, do not make claims about the brain, therefore neuroimaging evidence cannot be used to support one cognitive theory over another.’ Discuss. (2018)

EITHER
(a) ‘Double dissociation evidence is superior because it does not rely on assumptions.’ Discuss.
OR
(b) Critically evaluate the extent to which neuroimaging can shed light on cognition. (2016)

EITHER
(a) Which (if either) kind of evidence has greater value for choosing between competing cognitive-level theories, behavioural evidence or neuroimaging evidence?
OR
(b) Discuss, with examples, whether – and, if so, how – implemented computational models can contribute to human cognitive psychology. (2015)

DELUSIONS

TOP

ESSAY QUESTION

What is the most promising theoretical account of monothematic delusions, such as those of sufferers of Capgras syndrome?

CORE READING

FURTHER READING

Maher (1974) was an important starting point for understanding monothematic delusions, arguing that they are normal responses to abnormal experiences—experiences that are themselves the products of neuropsychological deficits. Recent developments of the multi-factor approach of Davies et al. have focused on Bayesian methods; Davies and Egan (2013) is a good overview. Another issue to think about is whether delusions are beliefs. Currie (2000) argues that they are not. For criticism, see Bayne and Pacherie (2005). See also Tumulty (2011) and Hohwy and Rajan (2012). For an accessible survey of recent work on delusions, discussing both sets of issues, see Bortoletti and Miyazono (2015).

PAST PAPER QUESTIONS

Do delusions involve abnormal cognition, in addition to abnormal experience? (2020)

What role, if any, do abnormal experiences play in the formation and persistence of delusional beliefs? (2019)

How many factors are involved in monothematic delusions, and what are they? (2018)

How strong is the case for a multi-factor account of Capgras delusion? (2017)

UNDERSTANDING MINDS

TOP

ESSAY QUESTION

How, if at all, do Simulation theories of understanding other minds differ from Theory theories? If they do differ, is there any evidence that decisively favours one over the other?

CORE READING

Gopnik and Wellman (1992) and Stich and Nichols (1992) are reprinted in Martin Davies and Tim Stone, eds. (1995) Folk Psychology: The Theory of Mind Debate (Blackwell) alongside various other classic papers on the topic and a helpful introduction written by the editors.

FURTHER READING

Ravenscroft (1997/2016) is an SEP entry on Theory theories. Gallese and Goldman (1998) suggest that simulation theory gains support from the discovery of mirror neurons in monkeys. Saxe (2005) assesses this in the context of the argument from error, and finds it wanting. Short and Riggs (2016) respond to Saxe, defending the simulation theory.

PAST PAPER QUESTIONS

‘Simulation theory and theory-theory are both required to explain how we read the minds of others.’ Discuss. (2020)

Is there good empirical evidence to suggest that we understand others through simulation? If there was, would that tell against the hypothesis that we exploit a theory of mind in understanding others? (2019)

EITHER
(a) What is the best theory of how we make sense of other minds?
OR
(b) Is the question of how we make sense of other minds amenable to empirical study? (2018)

EITHER
(a) ‘In making sense of other minds, simulation theory and theory theory really come to the same thing.’ Discuss.
OR
(b) ‘People make incorrect predictions about the mental states of other people who are just like them. The theory theory allows a ready explanation of such errors; the simulation theory allows none.’ Discuss. (2017)

EVOLUTIONARY PSYCHOLOGY

TOP

ESSAY QUESTION

What, if anything, can evolutionary arguments tells us about the mind?

CORE READING

FURTHER READING

The rest of Sterelny and Griffiths (1999) provides an excellent introduction to the background issues in the philosophy of biology, and is well worth reading. It is useful too to get a sense of alternative accounts of the Wason card selection test, starting with Sperber et al. (1995). (Although sceptical of Cosmides and Tooby’s account of the test, Sperber himself is quite sympathetic to evolutionary psychology, and is a proponent of the massive modularity thesis.) Lloyd (1999) argues that Cosmides and Tooby’s claims rest on a faulty understanding of evolutionary biology. Dupré (2012) takes a similar line. Cosmides and Tooby (2013) is a more recent summary of their work. Downes (2018) is a useful survey piece.

PAST PAPER QUESTIONS

‘Evolutionary arguments about cognition ignore the fact that the mind is just a jumble of assorted fixes and tricks, not the product of an engineering design.’ Discuss. (2020)

Is there room for evolutionary arguments in cognitive science? (2019)

How convincing are evolutionary explanations of particular aspects of cognition? (2018)

Do any evolutionary arguments have explanatory value in the cognitive sciences? (2017)

FREE WILL

TOP

ESSAY QUESTION

What, if anything, do Libet’s experiments tell us about free will and moral responsibility?

CORE READING

FURTHER READING

Daniel Wegner claims that Libet’s experiments show that free will is an illusion, albeit an important one. See his (2004), a précis of his book, The Illusion of Conscious Will. It’s followed by various critical responses. (Libet himself disagreed with Wegner, and thought that, while the experiments show that brain activities that lead to action begin before the agent is aware that she intends to act, the agent still has the capacity to veto the act in question.) Dennett (2003) is a lively response to Wegner. Mele (2009) sets out his criticisms in more detail. Bayne (2011) and Nahmias (2014) are two other good responses.

PAST PAPER QUESTIONS

Do the Libet experiments establish that free will is an illusion? (2020)

Is there any good empirical evidence from cognitive science that we do not have free will? (2019)

EITHER
(a) Is there an argument that we are not responsible for our actions to be made from Libet’s experiments?
OR
(b) Explain the role of the phenomenology of agency in our mental lives. (2018)

How strong is Libet’s argument that we are not responsible for our actions? (2017)

CONCEPTS

TOP

Coming soon.

PAST PAPER QUESTIONS

Are concepts structured? (2020)

Are philosophers and psychologists talking about the same thing when they use the word ‘concept’? (2019)

‘No single cognitive competence underlies our various capacities for categorization, learning, and reasoning. Therefore, there is no useful notion of concept for the purposes of psychology.’ Discuss. (2015)

INNATENESS

TOP

Coming soon.

PAST PAPER QUESTIONS

‘Treating a cognitive capacity as innate rather than acquired adds nothing to the explanatory power of a psychological theory.’ Discuss. (2019)

What role, if any, do innate components play in cognition? (2018)

How good is the case for innate aspects of cognition? (2017)

‘Children are born as blank slates; there is nothing innate in cognition.’ Discuss. (2016)

THE FRAME PROBLEM

TOP

Coming soon.

PAST PAPER QUESTIONS

‘The frame problem shows that cognition cannot be formalised via a finite set of basic axioms, and this shows that cognitive science is doomed to failure.’ Discuss. (2020)

Is the frame problem merely a problem for the study of artificial intelligence? (2019)

Describe the frame problem, and discuss whether it is solvable. (2018)

What is the frame-problem a problem for? Can it be overcome? (2016)

PSYCHOLOGICAL EXPLANATION

TOP

Coming soon.

PAST PAPER QUESTIONS

What are the differences, if any, between functional and mechanistic types of explanation in cognitive science? (2020)

EITHER
(a) What are Marr’s three levels? Can we fruitfully study the mind in any one of these levels independently of the other two?
OR
(b) What would it be to explain a cognitive capacity in terms of functional decomposition? How would such explanations differ from explanations in terms of laws? (2019)

EITHER
(a) How do Marr’s levels relate to each other?
OR
(b) ‘The personal/subpersonal distinction is best understood as a distinction between two types of psychological explanation – “horizontal” and “vertical”.’ (DRAYSON) Explain and discuss. (2017)

What are the distinctive features of cognitive scientific explanation? (2014)

• Home

Reading Lists

• Early Modern Philosophy
• Knowledge and Reality
• Ethics
• Mind
• Logic and Language
• Mathematics
• Cognitive Science
• Philosophical Logic

Lectures

• Frege and Tarski, HT 2018
• Philosophy of Logic, HT 2017
• General Philosophy, HT 2016