306 test 3

core concept of functionalism

Mental state and process types are defined by their causal-functional role, not by what physical substance they are made of

how functionalism differs from IT

Smart: Mental states are strictly identical to specific physical/brain-process types (e.g., pain = brain-process BP273).

Functionalism: Mental-state/process types are functional-state/process types which may be differently realized in different creatures

• functionally individuated: a semantic value, concept, or mental state that is defined and distinguished entirely by the causal or computational role it plays within a system, rather than by its physical makeup or external reference

token physicalism

all realizations of mental state types are physical realizations, even if types are functionally defined

analogy of internal combustion engine

We infer the existence of an engine's parts (carburettor, cylinders, spark plugs) based on input-output relations (fuel in, motion out). However, this functional organization gives no indication of what physical material the engine is built from.

cross species sharing

Arguments Against Identity Theory (Putnam)

Identity theory cannot easily account for sharing mental states with non-carbon-based life forms, dolphins, or dogs because physical makeups differ.

rylean functionalism

Rooted in Gilbert Ryle's logical behaviorism, but updated in light of objections

Causal roles include relations to both observable behaviors/inputs and other mental states

The ordinary ways of thinking about how people in particular mental states are disposed to behave provide us with the necessary resources for defining the different causal roles.

• 1/ MD is about genuine internal states of systems insofar as these states play particular causal roles in generating that system's manifest behaviour. ‘Real’ dispositions.

• 2/ The 'causal role' of a given mental state can include causal relations in which it stands to other mental states.

Thus, in characterising the causal-functional role of different mental states/processes, the inputs are observable states of affairs or other mental states/processes, while the outputs are observable behaviours or other mental states/processes

psychofunctionalism

methodological autonomy

A mature cognitive psychology will determine the overall functional organisation of a given cognitive system

This is to be determined on the basis of the observable discriminative and behavioural capacities of the system.

The 'inputs' and 'outputs' to the system will be given by other elements in our overall psychological theory for the system

methodological autonomy

the psychofunctionalist claim that cognitive psychology has a distinctive domain and laws discoverable independently of progress in neuroscience, similar to how an engine design is separate from the physical materials used to build it.

There is a distinctive domain of inquiry for cognitive psychology, and distinctive laws discoverable through inquiry into that domain.

• These laws are discoverable independently of progress in other sciences, such as neuroscience.

• This is so because the tasks of psychology and neurophysiology in respect of the behaviour of intelligent systems are distinct, though not completely unrelated

turing machine analogy

Minds can be understood as running machine tables where internal states are defined entirely by their causal-functional roles (inputs, outputs, and state transitions)

computational theory of mind

Complex cognitive abilities are explained through functional decomposition—breaking a system down into simpler parts and organizing their coordination

inverted spectrum problem objection

If two people experience different qualitative states (e.g., looking red vs. looking green) while their causal-functional roles stay identical (e.g., stopping at traffic lights), functionalism would fail to distinguish them

qualitative experiences like looking red vs green can invert while funcitonal roles remain identical, proving mental states cannot be purely identified with functional state types

inverted spectrum problem response

The qualitative features of the mental states change; their causal-functional roles don’t

qualitative states can accompany the same functional profile

absent qualia problem (objection)

If a billion people in China are linked via technology to simulate the exact functional organization of a human brain, functionalism dictates that the entire system must possess a conscious mind and qualitative experiences (like pain or a feverish feeling)

Because this system lacks actual qualia, functionalism appears false

the objection that a system can realize the exact functional organization of a human mind without having any conscious qualitative experience

chauvinism

Defining mental states too strictly can exclude systems (like creatures with different memory or computational structures) that intuitively share mental states with us

a flaw where a theory fails to ascribe mental states to creatures that surely do or could have them due to different functional organization

liberalism

Defining inputs/outputs too broadly (or poorly specifying them) can mistakenly ascribe mental states to non-thinking objects (like fairground games or coke machines)

problem/flaw where a theory of mind incorrectly ascribes mental states to entities that surely don’t have them

big question

how, in general, the different functional roles are to be characterised — who determines what the causal-functional role identifiable with pain-states is?

Block distinguishing two kinds of functionalism according to how this question is answered:

Functionalism (or Rylean Functionalism).

Psychofunctionalism

psychofunctionalism inputs/outputs

The inputs may be from various perceptual modules that act as transducers of information received from the environment, providing some kind of symbolic representation of that information to the system.

The outputs may be signals sent to various motor modules that are responsible for producing various behaviours of the system.

this implies that psychology is methodologically autonomous with respect to other sciences that investigate the behaviour of intelligent systems.

methodological autonomy task of cognitive psychology

to determine the functional organisation of intelligent systems, and to formulate laws that apply to a system in virtue of its functional organisation

methodological autonomy task of cognitive neuroscience

to examine particular realizations of a given model of functional organisation - e.g. the way that functional organisation is realized in humans, which may be distinct from the way it is realized in dolphins.

The neuroscientist can then appeal to more general laws in neuroscience to explain how the realization works

cognitive psychologist

we have, as data, a complex set of manifest discriminative and behavioural capacities.

We hypothesise a particular functional organisation that would enable the system to do what it does.

We test these hypotheses by, e.g., running a computer simulation of a system with the proposed FO and seeing how closely its behaviour resembles that of the system in question.

This task is constrained by neurophysiological considerations in the following respects: the functional organisation proposed must be realisable in the system in question, given its physical make-up.

• 1/ E.g. suppose it is claimed that we are able to compute a potentially infinite number of instances of a given mathematical function (e.g. addition).

• 2/ Might argue that this cannot be explained by proposing that we have stored an infinite number of equations, since the brain only has a finite storage capacity.

• 3/ So we must posit some kind of computational capacity, recursively employing certain rules or procedures.

e/ Where we have competing models of a system’s FO, we may be able to arbitrate between them when we know more about the physical make-up of the system.

cognitive neuroscientist

will also be guided by the cognitive psychologist in her efforts to understand the behaviour of the system:

The task for cognitive neuroscience is to identify the physical mechanisms that play the required functional roles - identify the relevant neuro-physiological sub-systems.

the crucial claim is that, insofar as neuroscientists are interested in explaining the behaviours of intelligent systems, they must individuate neurophysiological sub-systems by reference to their functional roles, and it is the cognitive psychologist's task to determine the latter

churchland

Why not be satisfied with intra-specific identities?

If our interest is in, e.g. pain-in-humans, then we cannot abstract from the work of the neuro-physiologist.

So psychology is not methodologically autonomous.

Fodor and Putnam responding to Churchland

It isn't clear that there are even species-specific identities, since we have neuronal plasticity - a given cognitive capacity can be realised by (neurophysiologically) distinct neural systems in humans - e.g. the recovery of certain linguistic abilities after neurological trauma.

Even if there are species-specific identities in the case of sensations and emotions, this is much less plausible when we turn to beliefs and desires.

Even if there are species-specific identities, the generalisations of interest to psychology may be cross-specific, and we will only discover such generalisations if we proceed autonomously in cognitive psychology.

turing

Alan Turing was a mathematician interested in the question: which mathematical problems can be resolved by means of some specifiable procedure which is guaranteed to yield a determinate answer in a finite number of steps?

Talk, here, of there being an algorithm for solving the problem, or of the problem being effectively decidable.

effectively decidable problems are ones that could be answered by a particular kind of computational device/ Turing machine:

a device whose operations could be completely described in terms of a set of functions defined upon inputs, outputs, and states of the system:

These functions fall into two types:

• a/ F(input, state) -> output

• b/ F(input, state[t]) -> state[t+1]

If we have a complete set of such functions, defined over all input/state pairs for the system, we can represent these functions by a machine table

This table provides us with an abstract characterisation of a device, and this characterisation can be realized by different physical mechanisms

fodor coke machine example

illustrates that states in a functional model or Turing Machine are defined entirely by their operational rules—how specific inputs cause particular outputs and state transitions—rather than by their physical construction

In the vending machine model, the internal states S1 ("needing a dime") and S2 ("needing a nickel") are specified strictly by their causal roles in processing nickels and dimes to dispense coke and make change

Consequently, proposing such a state system serves as an empirical hypothesis to explain observed behavior, remaining physically neutral so that the same functional organization could be realized equally well by mechanical hardware or by a human inside the machine.

functionalism appealing to turing machines

2 different contexts

The 'mind-body' problem tries to understand what minds are, or what it is to have a mind, or to have the various mental states that we ascribe to twm's.

Mental states are analogous to the states of a TM. For a physical system to have a mind is analogous to a physical system realizing a TM.

Also, TM's are devices, realisable in physical systems, that are capable of executing very complex tasks through sequentially performing simple tasks. Minds can be understood as working in the way that TM's work.

fodor computational theory of mind

The capacities of various systems, whether or not we think they have minds, can be given an explanation in terms of functional decomposition, where complex abilities of the system as a whole are explained in terms of:

• Decomposition of the system into parts that have simpler capacities.

• Model of how the operations of the parts are organised and coordinated. [See model of internal combustion engine].

One could accept a functional explanation of how minds work while rejecting a functional explanation of what minds are - e.g. Smart could accept this.

On the other hand, if one does offer a psychofunctionalist theory of what minds are, then a 'functional decomposition' explanation of mental capacities is one way of responding to the charge that functional explanations are vacuous as explanations -

• 1/ e.g. Q: How does a CD player know how to find track six? A: It contains a track-counter.

• 2/ More generally, we explain how a system is able to do A by ascribing to the system an A-causer.

• 3/ 'Humunculus' problem, here, for functionalist explanations of mental capacities - e.g. how are we able to visually discriminate things in our environment?

putnam evidence

if our evidence for ascribing particular kinds of mental states to organisms is similarities in their patterns of behaviour, this is good evidence to infer to the existence of things playing given functional roles, but not to the physical make-up of the things playing those roles:

We infer to the existence of an internal combustion engine from certain input-output relations (input - fuel, oxygen; output - motion). So we assume there are elements playing the causal-functional roles of fuel storage tank, fuel pump, carburettor or equivalent, cylinder, spark plugs, crankshaft, camshaft, etc. But this is no indication that the things that play these roles are made of one physical substance rather than another.

This suggests that, if the meaning of our talk of mental states is to be understood in terms of states playing particular causal-functional roles, then what we mean by e.g. experiencing pain is not the occurrence of a given type of process playing CF role (pain) in us (or some particular creatures); but the occurrence of the kind of process that plays CF role (pain) in the kind of creature to whom pain is being ascribed.

putnam long

the IT cannot account for the possibility that we share various mental states/processes with other animals, or with other life forms. If such sharing is to be possible, we must be able to specify a type of physico-chemical state/process so that all and only creatures in that physical state/undergoing that physical process are in a given mental state/undergoing a given mental process. Further, the correlation must be nomological, and apply to non-carbon-based life forms.

Even if we could specify such a type of physico-chemical process for pain, the IT theorist is committed to there being such an identity for all psychological states/processes that could be shared.

smart long

Types of mental states/processes are types of physical state/processes individuated as particular mental states/processes in terms of their causal-functional role:

Neuro-physiology identifies particular types of brain-processes in terms of their neuro-physiological properties.

Sensation(s) turns out to be identical to one such process because it is that brain-process that plays causal-functional role C(s). ‘x is in pain’ means ‘process P is occurring in x, where P has CF role (pain)’

As a matter of empirical fact, the process that plays this causal role in us is a type of physical process, brain- process BP273. Thus pain-experiences are strictly identical to occurrences of BP273.

troubles with functionalism

To evaluate functionalism as a theory of mind, need to answer the following questions:

does it capture the class of things that (intuitively) have mental states? → liberalism, chauvinism

does it correctly ascribe mental states to such organisms? ['Inverted spectrum' problem]

smart’s view of mental states

mental state-types are physical state-types that are functionally individuated, meaning pain is strictly identical to a specific brain process

For Smart, a particular type of mental state or process turns out to be identical to a specific physical brain-process because it is that physical brain-process that plays a specific causal-functional role in us (e.g., pain-experiences are strictly identical to occurrences of a specific brain-process like BP273)

putnam’s argument against the IT

cross species sharing

IT cannot account for the possibility of sharing mental states with non carbon based forms of life or other animals that lack the exact same physico-chemical brain state

ICE analogy issue

behavioural evidence (input-output relations) lets us infer the existence of functional roles like a carburetor, but it gives no indication of the physical substance making them

functionalist view of mental states

multiple realizability

mental state-types are functional state-types that may be differently realized/implemented in different creatures

neurophysiological constraint on functional organization

any proposed functional organization must be physically realizable in the system given its physical makeup

neuronal plasticity objection to species-specific identities

fodor and putnam’s argument that a given cognitive capacity can be realized by neurophysiologically distinct neural systems even within the same human (ex recovery after trauma)

functionalist response to absent qualia for qualitative states

proposing that qualitative mental states are composite, requiring both a functional role and a qualitative physical state realization

device a vs device b temperature example

two systems sharing the same input-output capacities and beliefs through entirely different functional organizaitons (one computes via formulas, the other uses a lookup table)

Consider two devices which are able to provide equivalences in F and C for all temperatures in whole numbers from the freezing point to the boiling point of water. Each will accept as

input C[0] - C[100], and F[32] - F[212]. Each gives as output the equivalent temperature to two decimal places.

We might say, of each device, that it ‘believes’ that F(70) = C(21.11)

But the devices might have very different functional organisations:

Device A has a computational module which is able to compute the answers by applying the following rules and rounding to two decimal places:

Device B has a list of equivalences, to two decimal places, for each possible input, and a search function.

Couldn't there also be twm’s that, intuitively, share mental states with us, but do so in virtue of a different kind of functional organisation from ours, as with these machines? ‘Chauvinism’ charge here against functionalism.

input output specification challenge

defining inputs and outputs without being too liberal like a coke machine or fairground game or too chauvinist

Obvious chauvinism of the Rylean Functionalist here, if we talk of inputs as visual stimuli and outputs as limb movements (ex animals?)

Standard formulations of Psychofunctionalism face the same problem.

Block argues on that no physical specification of the inputs and outputs can get round this problem - we get something that is either too liberal or is chauvinist.

Psychofunctionalist could perhaps get round this if she can avoid any reference to neural activity and talk only about information received from transducers, and output to motor modules.

nagel bats

Nagel argues that an organism has conscious mental states "if and only if there is something that it is like to be that creature—something it is like for the creature".

Bats use echolocation (sonar) to navigate and perceive their environment. Even though humans can understand the physics or biology of how echolocation works, we cannot experience or imagine what it feels like to actually perceive the world through a bat's senses.

Objective science relies on viewpoints that are detached from any specific individual perspective. Because conscious experience is inherently tied to a specific point of view, physicalist theories struggle to explain or reduce the subjective nature of the mind