Sumários

The perception-action theory

18 Novembro 2025, 14:00 • David Yates

November 18: The Perception-Action Theory

In this class we consider some famous lesion studies carried out by Milner and Goodale, from which they conclude that the visual brain is divided into separate perception and actions streams, with the activity of the action (dorsal) stream largely unconscious, and visual consciousness localized somewhere in the perception (ventral) stream. We consider their findings relating to patient D. F. (a case of visual agnosia due to carbon monoxide poisoning), and consider whether they are entitled to draw substantive conclusions from this case about the neural correlates of visual experience in normal brains. We then consider how they use optical illusion studies of individuals with normal brains to support their case. 

Reading:

Milner, D. & Goodale, M. (2006) ‘One Brain, Two Visual Systems’ 

Questions:

  1. What do Milner and Goodale infer from the double dissociation between perception and action that we see in cases of visual form agnosia and optic ataxia? Are they right? Discuss the second question in the context of potential alternative explanations of the experimental data.
  2. What is the role of the Ebbinghaus illusion experiment in offering additional experimental support for the two streams hypothesis?
  3. According to Milner and Goodale, motor actions such as grasping an object and posting a letter can be handled by the unconscious "where" stream. Assume they are correct. What then is the function of the conscious "what" stream?
  4. What, if anything, do Libet's timing experiments show us about the nature of freewill?

Further Reading:

McIntosh & Schenk, (2009) 'Two visual streams for perception and action: current trends'

Whitwell, Milner & Goodale (2014). 'The two visual streams hypothesis: new challenges and insights from patient D.F.'

Haffenden & Goodale (1998). 'The effect of pictorial illusion on prehension and perception'

Benjamin Libet (1999). 'Do we have freewill?'

Trevena and Miller (2010). 'Brain preparation before a voluntary action: evidence against unconscious movement initiation'

Split Brains and the Unity of Consciousness

11 Novembro 2025, 14:00 • David Yates

November 11: Split Brain Studies

In this class we will look at some classic lesion studies involving "split brains" - that is, patients whose corpus callosum has been severed. As we will see, such patients have many profound implications for the way we think about ourselves and our minds. How many minds do we have? How many people are we? How many centres of consciousness do we have? Is the unity of mind and consciousness an illusion? Why do we confabulate? Are most or even all of the narratives we construct to explain our own behaviour fictitious?

Reading:

Thomas Nagel, "Brain Bisection and the Unity of Consciousness"

Michael Gazzaniga, "The Split Brain in Man"

Tim Bayne, "The Unity of Consciousness and The Split Brain Syndrome"


Study Questions:

  1. What, if anything, do split brain patients tell us about the minds of neurotypical individuals?
  2. How does Tim Bayne use the "switch model" to argue that consciousness is unified, even in split brain patients? Do you thinj Bayne's argument works?
  3. Why do split brain patients provide evidence of left brain "confabulation"? What is confabulation, and what is its significance for cognitive science?
  4. Is the concept of a single, unified mind an empty concept that will be eliminated from mature cognitive science?

Neural Correlates of Consciousness

4 Novembro 2025, 14:00 • David Yates

November 4: Neural Correlates of Consciousness

In this class we will introduce the search for the neural correlates of consciousness and study some of its main techniques and problems. We will first ook at some evidence for unconscious representation in the visual brain, particularly from optical illusions. We will then consider Chalmers’ definition of a neural correlate of consciousness in terms of content-matching. We will study the role of lesion studies in locating the neural correlates of visual consciousness, and examine the problems Chalmers raises with such studies. Finally, we will consider binocular rivalry and the challenge of how to separate the neural correlates of visual consciousness per se from the neural correlates of post-perceptual cognition and report.

Reading:

Chalmers, D. 'What is a neural correlate of consciousness?', in particular pp. 1-11 and 14-21 (but see also section 5, pp 23-27 for a concise summary of his main methodological conclusions).

Questions:

  1. What, according to Chalmers, is a NCC for a particular type of conscious experience?
  2. What is representationalism about consciousness, and how does it feature in the matching-content approach to NCCs? Do you think representationalism is true?
  3. What evidence is there for unconscious visual representation? Why do unconscious representations pose a problem for NCCs research?
  4. Why are lesion studies problematic when it comes to localizing the NCCs? How might these problems be solved?
  5. What is binocular rivalry, and how is it used to investigate the NCCs for visual consciousness? What are the main problems with this approach?

If you want to see some optical illusions:

http://www.michaelbach.de/ot/index.html [many interactive illusions to browse]

https://www.illusionsindex.org/ [includes discussion of significance of illusions]

Further reading:

SEP entry on the neuroscience of consciousness

Crick & Koch, 'Consciousness and Neuroscience'

Block, 'What is wrong with the no-report paradigm and how to fix it'

And finally, an example of the masking technique I talked about as a way of solving the problem of verbal reports with bilocular rivalry studies:

Melloni et. al. Synchronization of Neural Activity across Cortical Areas Correlates with Conscious Perception

Connectionism

28 Outubro 2025, 14:00 • David Yates

October 28: Connectionism

Following on from the previous lecture, in this class we will continue to examine connectionism and consider how a connectionist network is able to solve the problems raised by Fodor: if cognition is isotropic and Quinean, how can we implement it computationally? We will consider an example of a connectionist network (Ramsey, Stich & Garon) to illustrate the differences between connectionist and classical function mapping. We will then use ChatGPT as an example to illustrate why connectionist networks are naturally isotropic and (at least in principle) Quinean. We will also consider the nature of representation in connectionist networks: if human cognition is connectionist, do we really have states such as beliefs and desires? Or is folk psychology a radically mistaken theory?

Reading:

T. Horgan, Connectionism and the Philosophical Foundations of Cognitive Science

See also: Churchland, P. 'Eliminative Materialism and the Propositional Attitudes'

For details of an early connectionist network that seems to lack any kind of LOT system of representation, see Ramsey, Stich & Garon 'Connectionism, Eliminativism, and the Future of Folk Psychology' (source of the technical slides discussed in the lecture)

Questions:

  1. How are connectionist networks able in principle to solve the problems of isotropic and Quinean cognition? Do they solve those problems completely?
  2. How can a connectionist explain the results of Wason selection tasks? What are the advantages of connectionist over classical architectures in this regard?
  3. What, if anything, does ChatGPT know?
  4. If the architecture of the mind is connectionist, and there are no language-like representations in the brain, do you really have propositional attitudes such as the belief that you are now reading this sentence?

Other relevant readings:

For more on the details of ChatGPT, see: https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/

For a comprehensive survey of arguments for and against eliminativism, see SEP

Cognitive holism

21 Outubro 2025, 14:00 • David Yates

October 21: Cognitive Holism, Modularity and Connectionism

Please be careful to distinguish cognitive holism from semantic or meaning holism, which we discussed in connection with conceptual role semantics! They are completely different! Cognitive holism refers to the properties of being isotropic and Quinean, which we will define in class.

In this class we will examine the main alternative to the classical computational theory of cognition, connectionism. We will begin with several related problems for classical computational accounts of mind, which have nothing to do with how the symbols get their meaning. These problems focus instead on whether it is plausible that central cognitive systems, i.e. those that are not modular (specialised for a particular task in the way that e.g. vision is) are classical symbol processors. We will begin with cases in which cognition seems to be context-dependent in a way that classical computationalism rules out. Then, as we will see, Fodor himself gave some powerful arguments (from cognitive holism and globality) that central cognition cannot be classically computational. These problems can all be seen to motivate connectionist accounts of the mind. We will focus on the central differences between classical and connectionist cognitive architectures by studying a simple connectionist network, and the ways in which both cognition and representation differ between LOT and connectionism. We will then examine the way in which modern deep learning models of cognition, the currently dominant paradigm, developed from connectionism. 

Reading:

T. Horgan, Connectionism and the Philosophical Foundations of Cognitive Science. (This is a good and fairly easy to read summary paper that discusses the differences between classical and connectionist theories of cognition, and explains in a simple way Fodor's main arguments from cognitive holism (section 2).

Further reading:

Samuels, R. 'Classical Computationalism and the Many Problems of Cognitive Relevance' (esp. sections 4 and 5)

See also: Ludwig, K. & Schneider, S. 'Fodor's Challenge to the Classical Computational Theory of Mind'

And finally, if you want to read a little of the original: Fodor, J. Precis of The Modularity of Mind (with peer commentary)

Questions:

  1. What are the main differences between classical symbol processing and connectionist cognitive architectures? (Horgan, parts 1-3)
  2. What is it for cognition to be (i) isotropic, (ii) Quinean? Why do these properties pose a problem for the classical computational theory of mind?
  3. Is Fodor correct that the computational theory of mind is probably false for central cognitive processes such as belief-fixation (deciding what it is you believe, whether a given proposition is true, etc.)?
  4. What, if anything, do Wason selection tasks tell us about the nature of cognition?