Mental states are defined by their functional roles, not their physical implementation
Functionalism
The Core Idea
Mental states are individuated by their causal-functional roles - the relations between inputs, other mental states, and outputs.
To be in pain is not to have a particular brain state. Itâs to be in a state that:
- Is typically caused by tissue damage (input)
- Causes beliefs like âIâm hurtâ and desires like âmake it stopâ (internal role)
- Causes behaviors like withdrawal and saying âouchâ (output)
What matters is the role, not the physical realization.
The Functional Role
A mental stateâs identity = its position in a causal network:
Inputs (perception, environment)
â
Mental State M
â
Other mental states (beliefs, desires, etc.)
â
Outputs (behavior, action)
âBelief that itâs rainingâ = state caused by seeing rain, that causes taking umbrella, combines with desire to stay dry to produce staying inside, etc.
Multiple Realizability
Key argument for functionalism:
The same functional role can be realized in different physical substrates:
- Pain in humans: C-fiber firing
- Pain in octopuses: Different neural organization
- Pain in aliens: Silicon-based circuitry?
- Pain in computers: Electronic states?
If mental states were identical to brain states, octopuses couldnât feel pain (different brains). But they seem to. Therefore, mental states = functional roles, not brain states.
Why This Matters
For Cognitive Science
Functionalism licenses studying the mind computationally without caring about implementation:
- Software/hardware distinction
- Same algorithm, different machines
- Computational theory of mind
For Psychology
Mental states are causes of behavior. Functionalism explains how: through their functional role connecting stimuli to responses.
Against Type Identity
Mind â brain. Mental kinds â neural kinds. Psychology is autonomous from neuroscience.
For AI
If functionalism is true, anything with the right functional organization could think - silicon, biological, or otherwise.
Versions of Functionalism
Machine Functionalism (Putnam)
Mental states are like computational states in a Turing machine. Defined by state transitions.
Analogy: Mind is to brain as software is to hardware.
Analytic Functionalism (Lewis)
Mental state concepts are implicitly defined by folk psychology. âPainâ means whatever plays the pain role in our common-sense theory.
Psychofunctionalism (Fodor)
Mental states are defined by their role in the best scientific psychology, not folk psychology.
Application to Research
Computational Modeling
Functionalism justifies:
- Studying algorithms independent of implementation
- Treating brain as implementing computations
- Testing theories via computer simulation
If functionalism is true, a good computational model doesnât just simulate cognition - it instantiates the same mental states.
Cross-Species Comparison
Can study cognition across species despite different brains:
- Do octopuses have beliefs?
- Do bees have episodic memory?
- Questions about functional role, not neural substrate
Level of Explanation
Multiple levels are all legitimate:
- Neural: How itâs implemented
- Computational: What function is computed
- Algorithmic: How the function is computed
Marrâs levels - functionalism supports their independence.
Limitations and Critiques
Qualia Problem
Does functional role capture subjective experience?
- Inverted spectrum: Same functional role, different qualia (I see red where you see green)
- Zombies: Same functional role, no consciousness
- Maybe phenomenal properties arenât functional?
Liberalism Problem
Too permissive - anything with the right causal structure counts as having mental states:
- Does a nation have a group mind?
- Does a system of water pipes with the right causal structure feel pain?
- Intuition: somethingâs missing (Searleâs Chinese Room)
Narrow Content Problem
Functional role is in the head, but content (intentionality) might depend on environment (Twin Earth). Can functionalism handle wide content?
Problem of Other Minds
If mental states are just functional roles, and we can only observe behavior, can we ever know others have mental states? (Behavioral similarity doesnât prove functional role identity)
Connection to My Work
This framework shapes:
- Computational approach: Models capture functional organization, not just correlations
- Cross-linguistic comparison: Same cognitive function, different linguistic realization
- Level of analysis: When to explain algorithmically vs. implementationally
- Cognitive architecture: Functional components (working memory, attention) vs. neural localization
Examples:
- Language processing: Study algorithmic level (parsing, composition) without waiting for neuroscience
- Bilingual architecture: Same functional components (lexicon, syntax), different organization?
- Code-switching: Functional role of switching mechanism, multiply realizable across individuals
- Executive function: Defined functionally (cognitive control), measured behaviorally, implemented neurally
Relation to Other Frameworks
- Intentionality: Functional role might determine intentional content (functional role semantics)
- Embodied Cognition: Challenge to functionalism - maybe body matters, not just functional role
- Extended Mind: If functional roles can be realized anywhere, can they extend into environment?
- Methodological Individualism: Functionalism supports individual-level explanation (functional states in individuals)
Contemporary Developments
Minimal Functionalism
Accept multiple realizability, reject strong computational assumptions. Mental states are functional, but not necessarily computational.
Embedded/Situated Cognition
Functional roles include environment, not just internal states. Challenge: still functionalism or moving beyond?
Mechanistic Explanation
Replace pure functionalism with mechanistic accounts: identify components, operations, organization. More constrained than âany realization will do.â
Key Sources
- Putnam, H. (1967). âPsychological Predicatesâ (later âThe Nature of Mental Statesâ)
- Fodor, J. (1968). Psychological Explanation
- Lewis, D. (1972). âPsychophysical and Theoretical Identificationsâ
- Block, N. (1978). âTroubles with Functionalismâ
- Shoemaker, S. (1984). Identity, Cause, and Mind