Sumários

Atomic theories of meaning

14 Outubro 2025, 14:00 • David Yates

October 14: Atomic Theories of Meaning

The problem of holism discussed in the previous lecture motivates atomistic theories of conceptual content (/meaning), according to which content is determined not by holistic inferential relations between concepts, but by relations between individual concepts and the things in the world that they represent. Such theories are often endorsed by proponents of compositional semantics and computational theories of cognition because meaning is determined atomically for each symbol. We begin with the crude causal theory of content, according to which a concept means what reliably causes it. We move on to consider the disjunction problem, also known as the problem of error: if a concept means whatever causes it, then it seems to be impossible to apply that concept incorrectly. We will then consider teleosemantics as a solution to this problem, a theory of meaning that explains the meanings of concepts in terms of their biological function.

General Reading:

Fodor, J. ´Meaning and the World Order´, ch. 4 of his Psychosemantics (focus on pp. 97-110).

Specific Reading on Teleosemantics:

Karen Neander, 'Teleological Theories of Mental Content: Can Darwin Solve the Problem of Intentionality?'

This paper is a bit long, but you can focus on sections 1-7 (and also 10 and 11 if you want, but we will cover them in class). The early sections summarise some of the material on meaning that we have already covered in class. 

For more detailed reading on these issues, see Neander's SEP entry on teleological theories of content.

Questions:

  1. What is the crude causal theory of content, and why does it have a problem accounting for misrepresentation (i.e. error)? Why is accounting for error important in a theory of meaning?
  2. How does teleosemantics solve the problem of error? Do you think meaning can be adequately explained in terms of biological function?
  3. Does teleosemantics entail that no artificial intelligence could ever have meaningful states?
  4. How do causal theories of content enable us to reply to Searle's Chinese Room argument? Does the reply (as Searle claims) accept that Strong AI is false? Or is CCTM + causal semantics close enough to count as a version of Strong AI?
  5. What is the significance of the "Swampman" objection to teleosemantics? Do you think the objection is serious?

Conceptual Role Semantics

7 Outubro 2025, 14:00 • David Yates

October 7: Conceptual Role Semantics

Here we will study one of the main theories of how concepts acquire their meaning: conceptual role semantics (CRS), the theory that the meaning of a concept is determined by the inferences that the possessor of the concept is prepared to draw using that concept. We will examine how CRS enables us to respond to Searle's Chinese Room argument by grounding the meanings of concepts in their computational roles. We will then introduce the problem of holism: if conceptual content is determined by total inferential role, how can two people ever be in the same mental state? Finally, we will consider whether the problem of holism can be solved by appealing to the analytic-synthetic distinction. 

Seminar Reading:

Fodor & Lepore, ´Why Meaning (Probably) Isn't Conceptual Role.

Focus on pp. 15-24 (sect. 1 and about half of sect. 2). Obviously, if you want to read it all, then do.

See also this fairly easy to read IEP entry on CRS

Questions:

  1. What role does conceptual (or functional) role semantics play in the Systems Reply to Searle's Chinese Room argument?
  2. What are Frege cases and how does conceptual role semantics handle them? 
  3. Why is conceptual role semantics holistic and why is holism a problem, according to Fodor and Lepore?
  4. What is the analytic-synthetic distinction and how does it help conceptual role semantics solve the problem of holism? Does the solution work?
  5. What is Fodor and Lepore's argument that inferential roles are not compositional? How does this lead them to reject functional role semantics? (A more difficult question, requires more reading.)

The Chinese Room

30 Setembro 2025, 14:00 • David Yates

September 30: Classical Computationalism and the Chinese Room

In this class we will first examine Fodor's Language of Thought Hypothesis and the way in which it builds on the classical computational theory outlined in the previous lecture. We will then examine the prospects for CCTM in the light of Searle's famous “Chinse Room” thought experiment. Searle attempts to show that running a formal program is not sufficient for understanding, by imagining a system that runs a certain kind of language program but which does not understand the responses it produces. If Searle is correct, then it seems that Strong AI (the claim that the mind is a formal syntax-driven computer program running on the “wetware” of the brain) is false.

Seminar Reading

Searle's original article, plus open peer commentaries (try to read the original article up to an including the systems and robot replies):

Searle, J. ‘Minds, Brains, and Programs’, Behavioural and Brain Sciences 3, pp. 417-457.

For a simplified version of the argument, see Searle's later article, ‘Is the brain’s mind a computer program?’ (easier to read than the original, contains all the important points plus a summary of the responses that were given to the original article and Searle’s thoughts on some of them).

See also SEP and IEP entries on the Chinese Room (IEP generally simpler, SEP aimed more at philosophy audience).

Questions:

  1. On the assumption that CCTM is true, does anything in the Chinese Room understand Chinese? If so, why? If not, why not?
  2. Does anything in the Chinese Room understand Chinese? If so, what is it? If not, what are the implications for strong AI?
  3. What is the "Systems Reply" to the Chinese Room argument? Does itsucceed in defending strong AI?
  4. What is the "Robot Reply" to the Chinese Room argument? Does it succeed in defending strong AI?
  5. Is Searle right to suppose that consciousness is a necessary condition for a system's symbols to have meaning?

September 23: The Classical Computational Theory of Mind (CCTM)

23 Setembro 2025, 14:00 • David Yates

In the class on September 16, I did not have time to start talking about classical computationalism, or the reasons for its historial significance. In this class, therefore, we will pick up where we left off last time and focus on CCTM's reductive explanation of some of the most important features of rationality as a reason for its initial popularity. In this class we will focus on the notion of a Turing Machine and show how to give a simple algorithm for a particular logical inference that we all know how to make: disjunctive syllogism. According to CCTM, the mind has the same computational architecure as a Turing Machine. Thus, we will see an example of two of Marr's three levels of analysis: computational and algorithmic. If there is time, we will also study Fodor's arguments for the Language of Thought hypothesis, and see how this hypothesis is a particular version of CCTM.

Relevant Reading:

Sections 1-3 of this Stanford Encyclopaedia entry on CCTM: https://plato.stanford.edu/entries/computational-mind/. 

See also: Fodor, J. "Why there still has to be a language of thought", appendix to Psychosemantics.

P. S. When Fodor writes about "Aunty" he is referring to proponents of (at the time) trendy new neural network architectures (then called "connectionism") that deny LOT. We will talk more about those later in the course!

Study Questions

  1. What is cognition according to CCTM?
  2. How does CCTM help us to explain how a purely mechanical device could be rational?
  3. What is the Langauge of Thought Hypothesis and what arguments does Fodor give in support of it?
  4. What is special about LOT as a version of CCTM and how does it differ from versions that posit computation only at the level of whole propositions?
  5. What role did developments in AI research play in the downfall of CCTM?

September 16: Introduction to Cognitive Science.

16 Setembro 2025, 14:00 • David Yates

In this class we will first introduce the multidisciplinary nature of cognitive science, and then briefly discuss the role of philosophy in it. We will then proceed to introduce three key features of mind to be studied over the coming sessions: mental representation, cognition, and consciousness. If there is time, we will also introduce the classical computational theory of mind and begin to study how it is able to provide a mechanistic account of (at least some components of) human rationality.

Study questions:

  1. What is the significance of Marr's three levels of analysis for cognitive science?
  2. What is the Representational Theory of Mind?
  3. Why must cognitive science be interdisciplinary?
  4. What role(s) does philosophy play in cognitive science research?
  5. What ways are there to test between two alternative cognitive models for e.g. catching a ball?

Relevant Reading

Thagard, P. An Introduction to Cognitive Science 2nd ed. MIT Press 2005, ch.1.