PHIL 255: Week 12

Representation

Announcements

The final exam is scheduled for Dec. 21, 9-11:30, AL 105.

• Much of this exam will be based on lectures. Web notes will cover only a small part of lecture material.

• The exam will be cumulative, in that the discussion of Churchland will build upon issues in Graham.

Essay topics are available; the essay is due Dec. 1.

Traditional analytic epistemology

Thoughts are propositional attitudes, where propositions are abstract entities, the meanings of sentences.

There are analytic sentences, true by virtue of meaning.

Thought experiments can reveal conceptual truths, known a priori.

Philosophical problems arise from conceptual confusion and can be eliminated by analysis.

Naturalistic alternatives

Thoughts are mental representations, which are neural structures that encode aspects of the world.

• Desires are not propositional attitudes.

There is no analytic-synthetic distinction: Quine.

Thought experiments may be useful for generating hypotheses, but have no probative value.

Philosophical progress is made by developing better explanatory theories of knowledge, reality, and morals.

How Brains Represent

Failure of behaviorism shows the need for postulating internal representations.

Brain representations are not primarily linguistic: continuity with children and other animals.

Neurons encode information.

• Local coding: one neuron encodes a property.
• Vector (population) coding: group of neurons encodes multiple properties.
• Encoding occurs by neuronal firing:
  • Average firing rate.
  • Pattern of spikes.

Neural representation:

• Occurrent: current pattern of activity in a network.
• Abeyant (stored): configurations of connection weights.
• Learning of connections by external stimuli or internal modeling.

Neurosemantics

Traditional view: syntax, semantics, pragmatics of language.

Fodor: innate language of thought.

Neural representations have meaning like maps do.

Activity in neural representations have causal correlations with objects in the world.

Inference

Traditional view: inference is deduction in a formal system, or a probabilistic variant.

Alternative: inference is transformation of neural representations.

Innateness

Traditional nature vs. nurture debates are pointless.genes + environment in interact with feedback.

See diagram in Churchland, p. 367.

Review questions for Final Exam

The exam questions will be fairly similar to ones below.

Short answer questions (you will have to answer 5 of 6, 5 marks each):

1. What are reductive explanations in science? How do they differ from correlations?
2. Why does Churchland think that reductionism is more plausible than functionalism or substance dualism?
3. How do neurological mechanisms support self-representational capabilities?
4. Do people have direct self-knowledge?
5. Why does Churchland think that defining consciousness is unnecessary?
6. What is the difference between the direct and indirect approach to experimental studies of consciousness?
7. Why does Churchland think that the zombie thought experiment does not rule out a neurological explanation of consciousness?
8. What is the inverted spectrum argument? Is it compelling?
9. Can uncaused actions be free?
10. How can a materialist distinguish free from unfree actions?
11. Are emotions compatible with free action?
12. How do neurophilosophical views of knowledge differ from those traditional in analytic philosophy?
13. List five mechanisms that enable brains to represent the world.

Short essay questions (you will have to answer 1 of 2, 10 marks):

1. What is the self? Compare two alternative views and argue in favor of one of them.
2. What is the strongest argument that consciousness is a neural process? What is the strongest argument that it is not? Evaluate these arguments and reach a conclusion.
3. Is free will consistent with neuroscience?
4. Can human knowledge be understood in terms of how brains represent the world?

Phil 255

Computational Epistemology Laboratory.

Paul Thagard

This page updated Nov. 28, 2005