Cognitive Science studies the mind as a biological information processing device, concentrating on core computational mind, or factory-installed representation-making and representation-using equipment. As such it draws from a number of fields -- neuroscience, psychology, linguistics, computer science, philosophy, etc. -- but only from the parts of those fields that are concerned with computation and representation. Cognitive science is to the study of mental life what evolutionary theory is to biology -- a unifying theory.

As a science, cognitive science seeks to explain the mind without invoking magical or mysterious phenomena. For example, dreaming may be explained by various Freudian theories of repression and sublimation, but this kind of account fails to be empirical and predictive. Dreaming is perhaps better understood as a kind of mental disk-maintenance operation. Or consciousness, thought of in the popular mind as a kind of mental spirit, is actually a consequence of the timing of the delivery of representations from different parts of the mind-brain: there is no single conscious area of the brain, and even the areas associated with volition appear to go into operation after the areas activated for their associated actions! Your conscious, willing self is thus an after-the-fact editor of what your brain has already done! You tell yourself what you did!

As a science, cognitive science presumes that the (mental) world is knowable (completely knowable?) and that precise, rigorous, and causal/predictive accounts (explanations) can be given of the (mental) world. Cognitive science is an empirical pursuit, seeking observable data (hence materialist). It seeks to state its account of mental life through the design of the organism. This design stance (Dennett) leads to what might be called the engineering condition of cognitive science: if you know what something is by design, you can then make one. (So how do you make a minimal computational mind?) Cognitive science adopts the usual methods of any science: observation, experiment, disciplined intuition (not so bad as might be thought), modeling, simulation, deduction form theory, and observation of breakdown.

Question: what kinds of study of mental life do not adhere to these conditions? Is moral theory, for example, a science? Is Freudian theory a science of the mind? Is too much made of the idea of science. Chomsky once said something like "any discipline that has to call itself a science probably isn't."

Following the conditions of science and concentrating on the mind as a computational-representational device, cognitive science is concerned with seven more particular issues:

1. Functionalism: mental content is understood in terms of its causal role in mental life. What role, e.g., does memory play in behavior? Putative mental content that cannot be given a causal role in mental life is eliminated. Functional systems are those that can be described in terms of the causal roles of their components. Functionalism, moreover, allows for multiple realizability: different hardware might functionally realize the same mind.

Is a thermostat a functional system? In a sense it is since it can be described by its goals and the causal states that achieve those goals. Is everything a functional system. Are humans then no different from thermostats? One answer: yes, and that's a good thing (see eliminativism). No: humans are best described as teleological functional systems: their mental states have a causal role because they are for something. A human has beliefs in order to do something. Is this a cheap way out of the issue, thus letting magical human characteristics in the back door? But does a thermostat have a metal strip that responds to temperature thresholds in order to regulate the heating/cooling system? Does a thermostat do what it does for a purpose?

2. Levels of explanation (Marr, Flanagan, and others)

a. Task (ecological): what is the organism doing?

b. Means (computational): what mechanisms/algorithms can account for what the organism is doing?

c. Physical (material/implementational): what is the hardware/wetware in which the algorithm is implemented? What is the material basis of the computation?

A thermostat regulates the heating/cooling system (task) by sending signals to the system as a consequence of thresholds of temperature (computational), and this threshold detection is implemented in a strip of a certain kind of metal inside the device (physical). Is the thermostat a (teleological) functional system? Note that other thermostats could be designed that achieve these tasks in similar or even quite deifferent ways at some level. Imagine a programmed thermostat with a set of internal representations of temperature states. This would have the same functional account and achieve the same tasks, but do so differently form the standpoint of its mechanisms and hardware. Would this kind of thermostat be more like a mind?

3. The characteristics and relationship of levels:

a. Supervenience: a higher level X supervenes on a lower level Y is there can be no change in X without a change in Y. Supervenience is a way of having dependence across independent levels without falling into dualism (the terribly unfashionable positing of distinct kinds of matter). The computational supervenes on the physical in that changes in algorithms require changes in the hardware.

c. Individualism: the relevant phenomena can be accounted for without reference to properties outside the organism. The explanation of computational and hardware/wetware states does not need to make reference to the external world to be accurate. One thus explains language acquisition, for instance, independently of the social and cultural context in which it occurs, just as one explains data storage in a machine without reference to what that data stands for (non-content-addressable memory). Note how the design stance and engineering condition bear on this matter: explanations by design can be done without refernce to factors outside of the inner design.

Question: is the ecological level necessarily nonindividualistic?

d. Reductionism/Eliminativism: if the content of one level can be restated in terms of the content of another, the first reduces to the second. We hope to reduce mental states to brain states, for instance. One might then claim, however, that such reductions necessitate the elimination of the states reduced: why talk about consciousness if there is no such thing? But how should we then talk? If I'm afraid, should I say, "My amygdala's acting up?"

4. Intentionality vs. Intelligence: how do processes that don't seem to be about anything (internal brain processes of computational intelligence) come to be about the world (intentional)? Is a thermostat intelligent but not intentional?

5. Language of Thought: what is the inner code into which the world is rendered? Rich and abstract or shallow and connected to the world?

6. Computation: what does it mean to compute? The manipulation of information by its form?

7. Architecture: what is the overall mental/computational format of the inner code? Domain specific modules or a multiply interconnected network? Or both, at different levels of explanation?