Jerry Fodor – Psychosemantics: Chapter 1 Notes

Jerry Fodor – Psychosemantics

Chapter 1 : Introduction : The Persistence of the Attitudes

Introduction

• Fodor begins with a quotation from A Midsummer’s Night Dream, act 3 scene 2,
involving Demetrius, Hermia and, in absentio, Lysander. Hermia has deduced,
from Lysander’s disappearance, that Demetrius, his rival in love, has killed him.
• Fodor shows how the deduction has gone. Hermia believes herself beloved of
Lysander because he has told her and, ceteris paribus, we can deduce how people
feel from what they say. So, Lysander wishes Hermia well; but, if so, he wouldn’t
voluntarily abandon her at night in a dangerous wood. Since he has in fact
deserted her, it must be involuntarily, so must plausibility have come to harm. The
prime suspect is his rival Demetrius as it is plausible that love-rivals do not wish
one another well and Hermia believes that Demetrius believes that Lysander alive
is an impediment to his aims. She also believes the plausible argument that if x
wants P, believes that not-P unless Q and that he can bring about Q, then he will,
ceteris paribus, attempt to do so. She further believes that those who attempt are
usually successful. So, Hermia suspects Demetrius of murdering Lysander and
we, the audience, following her largely unspoken chain of reasoning and sharing
many of her psychological beliefs, sympathise with her suspicions.
• In fact, the ceteris paribus clause has been sprung by a peripatetic sprite and the
chain of reasoning relying on the interconnected web of beliefs, desires and
actions that Hermia, Shakespeare and we share has gone wrong.
• Fodor points out, however, that while commonsense1 belief/desire psychology
may fail occasionally, it is (1) often right (2) deep and (3) indispensable. It has
come under philosophical attack and Fodor now addresses whether it is worth
saving by looking further at these three points in its favour.

1. How Often It Works

• Our relations with one another are mediated by applications of commonsense

psychology (henceforth known as “the theory”, and break down when its
predictions fail. Such failures are the stuff of theatre, but its ubiquitous success
goes unnoticed.
• The theory works so well, it goes unnoticed. I may agree to meet someone I’ve
never met before some months hence and thousands of miles away, and all go
well. Failure to keep the appointment is more likely to do with transportation
difficulties than failure of the theory. We cope with one another by its use better
than we do with less complex machines.
• The theory tells us how to infer people’s intentions from their utterances; not just
for our intimate friends but complete strangers; and not in the lab but in the field.
If we could perform the simpler function of predicting the weather as well, we’d
never get wet.
• What about ceteris paribus ? Some philosophers say the theory is trivially
incapable of disconfirmation, and so false or vacuous.

1
Note that this is a less pejorative term for “folk psychology”.

[email protected] Page 1 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

• If we consider the defeasibility of an ordinary utterance, such as a promise, it

holds true unless the speaker is lying, saying unintelligible words, talking in his
sleep, etc. We might view promises encumbered by these ceteris paribus clauses
as equivalent to saying “I’ll do this … unless I don’t”. Nothing can disconfirm it.
• While many philosophers are attracted by this sort of argument, it can’t be right
because commonsense psychological generalisations work most of the time and
enable us to make predictions, so cannot be empty.
• Fodor thinks that all the special sciences, other than basic physics, rely on such
ceteris paribus clauses. A meandering river may fail to erode its outside bank if it
freezes, dries up or the world comes to an end. Do the ceteris paribus clauses in
this case make geology false or vacuous ?
• No – because the geological generalisation is not of the form “P or not-P”. It is
complex matter how the special sciences can be both hedged and informative,
supportive of counterfactuals and exceptions; and this relates to why we have
special sciences and not just basic physics. The ceteris paribus clauses mean “ …
in any nomologically possible world where the operative generalisations are
satisfied”, and this is stronger than “P in any world where not not-P”. Hence,
commonsense psychology relies on ceteris paribus clauses no more than geology.
• However, is this similarity superficial ? Can we, as Davidson suggests, “perfect”
the generalisations of the special sciences by enumerating the conditions under
which their generalisations are supposed to hold ?
• Fodor thinks that, in this case, the only “real” science left would be basic physics.
His point is that to enumerate the exceptions to a geological generalisation, we
have to step outside geology. Explaining failure of erosion because of a change in
the weather is to invoke a meteorological event rather than a geological one. All
that can be said within geology is that, if the generalisation failed to be satisfied,
then the operative idealisations must somehow have failed to be satisfied. And so
with commonsense psychology; failure to turn up as announced indicates
something went wrong.
• Exceptions to generalisations are inexplicable in the vocabulary of a special
science, which is why it is special, but may be explicable in the vocabulary of
another science; most familiarly in terms of a more basic science. This is what the
hierarchical arrangement of the sciences provides. So, while we cannot discharge
the ceteris paribus clauses from within the special science, we have no reason to
doubt that we can do so in the vocabulary of some lower-level science (neurology,
biochemistry or, at worst, physics).
• The world is describable as a closed causal system, if at all, only in terms of our
most basic science, and this is irrelevant to psychology or geology.
• The moral of the above is that the predictive adequacy of commonsense
psychology is rationally indisputable and is not obtained by cheating. To find out
where I’ll be next Tuesday, it’s no use asking a physicist – you have to ask me !

2. The Depth of the Theory

• Fodor asks whether commonsense psychology is just a collection of Granny’s

nostrums, such as “money can’t buy happiness” and agrees that these are not
worth saving2. However, the bulk of commonsense psychology is non-
platitudinous and shows the deductive structure characteristic of real science.
2
See the end of the Introduction for Fodor’s aims in this paper.

[email protected] Page 2 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

There are two reasons for this (1) the theory’s generalisations are defined over
unobservables and (2) these generalisations lead to predictions via iteration and
interaction rather than direct instantiation3.
• Justifying (2) first - the example of Hermia shows that she’s relying on an implicit
theory of the interaction of beliefs, preferences and behaviours – ie. an implicit
decision theory. She’s no behaviourist, so thinks Demetrius’s behaviour is
intricately caused by his mental states. She doesn’t think that rivals necessarily
kill, ie. (x)(y)(xRy
xKy), but that the true and counterfactual-supporting
generalisation is that rivals prefer one another’s discomfiture. Behaviour can only
be deduced from this generalisation with the aim of assumptions about beliefs and
preferences.
• With respect to (1), it’s a deep fact about the world that the most powerful
etiological4 generalisations hold of unobservable causes, so this is a good test of a
deep theory. Fodor gives the example of commonsense meteorology, of the “red
sky at night” variety, which only involves instantiation and modus ponens, which
explains its inefficacy and shallowness. On the contrary, commonsense
psychology passes the depth test by assuming that behaviour comes at the end of
an arbitrarily long and tangled causal chain of unobservable mental events. We are
born and remain both mentalists and realists until persuaded out of it by bad
philosophy.

3. The Theory’s Indispensability

• We have no alternative to commonsense psychology if we want a theory of

counterfactual-supporting generalisations to explain our behaviour and its causes.
• Because commonsense psychology is so close to us, its value is often overlooked.
Commonsense psychological generalisations bridge the gap between exchanges of
utterances and subsequent behaviour. Without it we can’t even describe utterances
as words, or behaviour as acts. Word is an irreducibly psychological category5.
• Fodor specifies four conditions that require satisfaction if we want to avoid
commonsense psychology, and claims they cannot be satisfied. (1) Behaviour in
uttering x counts as an event of type Ti. (2) Fulfilling the import of utterance x
counts as an event of type Tj. (3) “Events of type Ti are consequent on events of
type Tj” is roughly true and counterfactually supporting. (4) Categories Ti and Tj
are not irreducibly psychological.
• Fodor’s point is that events of types Ti and Tj (utterances, statements, actions) are
psychological, so fail to satisfy (4).
• Philosophers and psychologists used to hope that the conceptual apparatus of
behaviour could be replaced by that of movements. As a result, psychological
generalisations would be seen to be the consequence of movements being
contingent on environmental and / or organic variables. While this contingency is
true, the generalisations don’t follow from it because behaviour consists of

3
Hopefully the examples will explain what this means !
4
What does this mean ?
5
With respect to the irreducibility, Fodor says this is “as far as anyone knows”. He focuses on the
acoustic properties of words, claiming that there are no acoustic properties that all and only tokens,
even fully intelligible tokens, of a particular word share, which explains the then difficulty of
producing working speech-recognition software. I’m not clear on the truth, importance or relevance of
this aside.

[email protected] Page 3 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

actions. Since actions cross-classify movements, we need to know the action type,
not the motion type6.
• It is generally assumed that the situation will be remedied in due course, as a
completed physics would subsume motions of everything, including organisms,
and eventually we will be able to have counterfactually-supporting generalisations
about the motion of organisms under the description of physics.
• Fodor thinks this is not only optimistic but misleading. A completed physics
would have no place for the behaver any more than behaviour. It would not
recognise organisms as such, but only matter.
• However, even if psychology were dispensable in principle, this is no argument
for dispensing with it, any more than it is sensible to dispense with rivers as a
natural kind, and with geology in favour of basic physics. Psychology would have
to be dispensable in fact, not just in principle, and it is clear that we cannot explain
ourselves other than in a vocabulary saturated with belief/desire psychology. We
can’t abandon the propositional attitudes because we don’t know how to.
• The rest of Fodor’s book is an attempt to vindicate commonsense psychology by
showing that, or better how, we can have a respectable science whose ontology
contains states with the properties attributed to the attitudes by common sense. He
now explores what properties these are.

The Essence of the Attitudes

• How do we know whether a psychology is of the belief-desire variety with the

propositional attitudes in its ontology ? Inter-theoretic identification –
distinguishing reduction from elimination from reconstruction is problematical;
even behaviourists were unsure whether they wanted to eliminate the mental or
equate it with the behavioural.
• Fodor stipulates that a psychology is common-sensical about the attitudes just in
case it postulates states, entities, events and so on that :-
i. Are semantically evaluable
ii. Have causal powers
iii. The generalisations of commonsense belief/desire psychology are true of
He takes these stipulations as intuitively plausible and squabbling about intuitions
as vulgar ! He now explains these stipulations.

(i) Semantic Evaluation

• Beliefs are true or false, desires can be fulfilled or frustrated and so on. According
to Fodor, what makes a belief true is its relation to the non-psychological world
rather than to other beliefs. Fodor terms the evaluation of the belief against the
world “semantic”.
• That beliefs, desires etc. are semantically evaluable is a puzzle as virtually nothing
else is. Propositions are, but this is unsurprising as propositions are what beliefs
and desires are attitudes towards. Semantic evaluability is the most
philosophically troublesome element of belief/desire psychology and the part
requiring most defence.

6
Fodor’s example is of the generalisation “the burnt child avoids the fire”, but the movement that
constitutes avoidance depends on a multitude of factors, such as the relative positions of the child and
the fire. I’m unclear what this is supposed to show.

[email protected] Page 4 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

• Content and semantic evaluability are intimately connected. If you know the
content of a belief, you know what it is about the world that determines its
semantic evaluation.
• Propositional attitudes have their content essentially7. To pick out an attitude we
need only to specify (1) the type (belief, desire, hunch ..) (2) the content (“2 is a
prime number”, …) of the attitude. Any propositional-attitude psychology must
individuate its states in this way.

(ii) Causal Powers

• Commonsense psychological explanation relies on three sorts of mental causation.

(1) Causation of behaviour by mental states, (2) proximal stimulation of mental
states by environmental events and (3) causation of mental states by one another.
The latter is the most interesting, giving chains of thought – a complex mental
event in which one semantically evaluable mental states gives rise to another until,
often, belief is fixed.
• All realist psychologies acknowledge the causal powers of mental states.
Functionalists hold that the identity of a mental state is determined by its causal
powers. Fodor doesn’t think this is correct, though it would be significant if it
were, for it would follow that the causal powers of a mental state determine its
content (given that propositional attitudes have their contents essentially).
• Any view vindicating belief/desire psychology attributes contents and causal
powers to the very same mental things it takes to be semantically evaluable.
• Further, causal relationships respect content relationships. For instance, if
someone believes Fa then, ceteris paribus, he believes (∃x)(Fx). Presumably, since
Fa ⊃ (∃x)(Fx) (by existential introduction), belief that Fa causes belief that
(∃x)(Fx)8.
• Fodor refers back to the introductory section, where Hermia believed the
generalisation “that if x wants P, believes that not-P unless Q and that he can bring
about Q, then he will, ceteris paribus, attempt to do so”. In this case, we can say
that we have explained x’s attempt to bring about Q given our knowledge of his
beliefs and desires about P and Q. What is typical of the explanation is (a) appeal
to causal relations between semantically evaluable mental states and (b) the
existence of content relationships amongst the mental states appealed to.
• Fodor points out that such explanations would fail if there were equivocation
about the schematic letters – it’s important that what’s believed can’t be had
unless (...) is what’s desired. The same contents often occur in causally related
mental states, with causal relationships respecting semantic ones.
• The similarity between trains of thought and arguments arises from the parallelism
between causal powers and contents. Fodor gives an example of Sherlock
Holmes’s deductive train of thought from The Speckled Band. The quotation
shows not just the psychology – the sequence of mental states that led Holmes to
suspect and then believe the method of the murder – but assembles premises that
are plausibly sufficient for a conclusion to be drawn by himself and others. What
connects the causal history aspects of the story to the plausible-inference aspects
is the fact that the thoughts that fix a belief that P are often grounds for believing

7
Look up essentialism !
8
Need to re-read this section to see whether I’ve missed anything subtle – eg. whether the causation
isn’t between mental states but between the world and mental states.

[email protected] Page 5 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

that P. If this – the general harmony between the semantical and causal properties
of thoughts - were not the case, there would be little point in thinking.
• Such thoughts raise a number of philosophical issues about the notion of content
that cause some philosophers to despair of commonsense psychology but which
need to be accounted for if a deep fact about the cognitive mind is not to be left
out. These questions include :-
1. What sorts of content relations are preserved in the generalisations9 that
subsume typical cases of belief/desire causation ?
2. How could the mind be constructed that such generalisations are true of it ?
3. What sort of mechanism could have states that are both semantically and
causally connected ?
4. How can causal connections respect semantic ones in such a mechanism ?

(iii) Generalisations Preserved

• Fodor considers, by way of summary, the properties required of a psychological

theory that vindicates commonsense belief/desire explanations. These must :-
1. Permit the assignment of content to causally efficacious mental states
2. Recognise behavioural explanations in which covering generalisations
quantify over the contents of mental states that they subsume
3. Not recognise generalisations that are crazy from the viewpoint of common
sense
4. Have attitudes with causal powers more or less as common sense supposes
5. Be at least approximately true.
• Fodor claims to be unphased by the fact that common sense is often mistaken
about the attitudes as about anything else. He hopes that there is more to the mind
than is dreamed of by common sense. Since Freud we have learnt that, contrary to
common sense, much of what’s in the mind in unconscious or unlearned. He
doesn’t have a list of commonsense generalisations that must be preserved by any
theory ontologically committed to bona fide propositional attitudes.
• However, we have no reason to doubt much of commonsense psychology. Any
psychology of action that is committed to the attitudes must acknowledge the
causal relations amongst beliefs, desires and behavioural intentions explicated by
decision theory. Similarly, any psycholinguistics which attributes beliefs, desires
and communicative intentions to speakers and hearers must entail an infinity of
theorems of the form “‘x’ is the form of words standardly used to communicate
the belief that x”.
• And again, any psycholinguistics that appeals to the propositional attitudes of
speakers and hearers to explain their verbal behaviour must entail that they know
at least one such theorem for each sentence ‘x’ in their language. Consequently,
psychology already has an infinite amount of common sense to vindicate10.
• Self-confident essentialism11 is currently fashionable and some philosophers have
modal intuitions about whether there could be cats in a world in which all

9
I need to check that I have given sufficient weight to generalisations in my account of Fodor.
10
Some of this argument must be tongue in cheek. What is it about ?
11
Essentialism : Fodor was writing in 1987. Essentialists maintain that some of an object’s properties
are essential to it, the rest are accidental. The essential properties of a thing are those it needs to possess
to be the thing it is. There are two forms of essentialism – definitional and modal. The latter has it that
x is essentially F iff necessarily whatever is x has the property F; alternatively, x must be F to exist at
all. See Kripke, Naming and Necessity for a de re modal epistemology (Concise Routledge

[email protected] Page 6 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

domestic felines were robots or Homer in a world where nobody wrote the Illiad
or Odyssey. Fodor doesn’t aspire to their enviable epistemic condition. He doesn’t
know how much commonsense psychology would have to be true for there to be
beliefs and desires, but he’s not worried as he has no doubts that lots of it is true.

RTM (Representational Theory of Mind)

• The main purpose of Fodor’s book is to argue that we have every reason to
believe that it is possible to have a scientific psychology that vindicates
commonsense belief/desire explanation. There already exists (in Fodor’s view) a
more or less empirical theory that is ontologically committed to the attitudes, is
quite probably approximately true, is a vindication of the attitudes and is the only
scientific belief/desire psychology in the field. Defending this theory and
defending the commonsense attitudes is the same thing.
• Fodor doesn’t think that recent arguments against belief/desire explanation are
cogent against this theory, which the rest of this chapter is briefly to explain.
• The theory is RTM (Representational Theory of Mind). The fundamental thesis of
the theory is that there is a language of thought, an infinite set of mental
representations which function both as the immediate objects of propositional
attitudes and as the domains12 of mental processes. Fodor says that RTM is made
up of two claims :-
• Claim 1 (The nature of propositional attitudes) : For any organism O, and any
attitude A towards the proposition P, there is a (computational/functional)
relation R and a mental representation MP such that MP means that P, and O
has A iff13 O bears R to MP.
• Claim 2 (The nature of mental processes) : Mental processes are causal
sequences of tokenings of mental representations.
• What Claim 1 means is that to believe that P is to have a mental symbol that
means that P tokened in your head; to have such a token in your “belief box”.
Similarly for hopes tokened in your “hope box”, the different boxes, shorthand for
functional states, reflecting the different causal roles of the attitudes.
• Similarly, Claim 2 means that a train of thought is a causal sequence of tokenings
of mental representations that express the propositional objects of the thoughts. To
have a train of thought “because x is the case I’ll do y” is to have a tokening of a
mental representation that means y caused by one that means x.
• Fodor’s reasons for thinking RTM more or less true are as follows. (1) Our best
science is our best guess at what there is and what it’s made of, and versions of
RTM underlie all current psychological research on mentation. (2) RTM suggests
a plausible mechanism for the parallelism between causal relations amongst
mental states and semantic relations between propositional objects, and the fact
that trains of thought are largely truth-preserving relies on this symmetry.
• The trick is to use the “computer metaphor”, since computers show how to
connect semantical with causal properties for symbols and therefore we can use
this as an analogy for thoughts if having a propositional attitude involves tokening
a symbol. This is the only major advance of contemporary cognitive science over

Encyclopaedia of Philosophy). Hence, returning to Fodor, if it is essential to being Homer that he wrote
the Illiad and the Odyssey, he would not exist in a world in which these works did not exist.
12
I need to express this in my own words when I understand better what he’s on about.
13
Fodor says that this biconditional is true strong, but not so as to undermine the spirit of the proposal.

[email protected] Page 7 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

18th and 19th century mentalism. Associationism failed because there was no way
for a rational mental life to arise from the causal relations between thoughts
recognised by the “laws of association”, as the closing pages of Ulysses illustrate /
parody.
• This is how the new story goes :-
1. The connection between the causal and semantic properties of a symbol is via
its syntax.
2. The syntax of a symbol is one of its higher order physical properties,
metaphorically, an abstract feature of its shape14.
3. The shape of a symbol is a potential determinant of its causal role, so we can
see how there could be environments where the causal role of a symbol
correlates with its syntax, where symbol tokens interact causally in virtue of
their syntactical structures.
4. The syntax of a symbol determines the causes and effects of its tokenings
much as the geometry of a key determines which locks it can open.
• Modern logic tells us that certain semantic relations amongst symbols can be
mimicked by their syntactical relations, which is ultimately what proof-theory is
about. Within famous15 limits, when there is an entailment relation between two
propositions, there will be a parallel between the semantic relation between the
two symbols and their syntactical relation whereby one symbol can be derived
from the other. Hence, we can build machines, again within the famous limits,
such that the machine :-
1. Operates solely by transforming symbols
2. Is sensitive solely to the syntactic properties of the symbols
3. Operates solely by altering the shapes of the symbols
Even so, the machine transforms one symbol into another only if the propositions
expressed by the symbols stand in relevant semantic relations to one another, eg.
of a valid argument between premises and conclusion. Such machines –
computers – just are environments in which the syntax of a symbol determines its
causal role by respecting its content.
• This connects with RTM’s ontological commitment to mental representations as
follows. Computers one means of mediating between the causal and semantic
properties of symbols. Hence, if the mind is a sort of computer, we can see how
we can get a theory of mental processes that succeeds, where all previous attempts
have failed, in explaining how there can be non-arbitrary content relations among
causally-related thoughts. For this to work, there need to be mental
representations. The parallelism evident in computers between the syntax and
semantics of symbols that brings causal role and content into phase won’t help the
theory of mind unless there are mental symbols possessing both semantical and
syntactical properties. We need mental symbols because only symbols have
syntax and our only theory of mental processes that isn’t known to be false
pictures the mind as a syntax-driven machine.
• Fodor now considers the objection of Churchland et al that commonsense
belief/desire psychology is a sterile theory that hasn’t advanced much since
Homer or at all since Jane Austin. He agrees that there is some truth in the charge,
as, though commonsense psychology may be implicit science, it isn’t research
science. Progress is to be found in the explicit vindication, where there has been

14
What does this mean ? See Fodor’s end-notes here and elsewhere.
15
Does this refer to Goedel’s theorems, or Turing’s halting theorems ?

[email protected] Page 8 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

enormous progress; and not just in our understanding of memory and perception
concerning the fixation of belief, nor about language concerning the
communication of belief. The real progress is that we are on the verge of
understanding how the causal processes of the mind can be semantically coherent;
in other words, how rationality is mechanically possible. Such problems can’t
even be posed unless we suppose, in accord with commonsense belief/desire
psychology, that there exist mental states with both semantic contents and causal
roles. A good theory invites questions that have answers.
• As was noted above, the biconditional in Claim 1 won’t do, and Fodor finishes the
chapter by polishing his claims for RTM.
• Claim 1 had it that both the following are true, whereas both are on occasion false
a) For each tokening of a propositional attitude, there is a tokening of a
corresponding relation between an organism and a mental representation.
b) For each tokening of that relation, there is a corresponding tokening of a
propositional attitude
• How does Fodor escape from this problem ? He makes an analogy with explicit
science co-opting the conceptual apparatus of common sense. Is water H2O ? Not
if the operative notion of water is the commonsense one that treats water as the
stuff we use in our daily lives. Only in the theory’s core cases of chemically pure
water is H2O, but “chemically pure” isn’t a commonsense notion. Still, common
sense is right in saying that there is water in the stuff we normally use and that
gives it the properties we rely on.
• What’s the relevance of this to RTM ? Fodor thinks that RTM vindicates
commonsense psychology in the cases that RTM identifies as core. In these cases,
what commonsense psychology takes as tokenings of propositional attitudes are
indeed tokenings of a relation between an organism and a mental representation.
The other cases, where the biconditional fails in either direction, are “derivative” –
and this is exactly what we’d expect from scientific precedent. Fodor now
considers a couple of examples in either direction both to see what the fuss is
about and to tighten up Claim 1.

Case 1 : Attitudes without Mental Representations

• This is from Dennett. A designer of a chess-playing program refers to a rival’s as

“wanting to get its queen out early”. Dennett points out that this is a usefully
predictive mental attitude, but that there is at no level in the program any explicit
representation tokened of this attitude. This attitude is an emergent property of the
computational processes that have “engineering reality”. Dennett thinks there’s no
reason to think that the relation between belief-talk and psychological-process talk
will be any more direct.
• Fodor starts of by pointing out that Dennett’s problem isn’t that some
propositional attitudes of commonsense psychology are dispositional. I may be
said to believe something I’ve never heard of just because I would believe it if I
were shown the proof. Dispositional beliefs couldn’t correspond to occurrent
tokenings of relations to mental representations. However, Claim 1 can be easily
patched up by saying that for a dispositional belief there is a corresponding
disposition to token a mental representation.
• This leaves open the question when are the attributions of dispositional beliefs
true ? According to Fodor, one’s dispositional beliefs should be identified with the
closure of one’s occurrent beliefs under principles of inference one currently

[email protected] Page 9 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

accepts. RTM can live with the uncertainty of such a closure as dispositional
attitudes play no causal role in actual mental processes. Therefore, RTM can
afford to be operationalist about dispositional beliefs provided it is firm about
occurrent ones.
• However, Dennett’s point was not that the chess-playing program had a potential
belief but an actual operational one, even there is no tokening of any symbol that
expresses it. Similarly, behavioural commitment to modus ponens might betoken
that this is inscribed in brain writing, but it needn’t as this rule might be complied
with rather than literally followed16.
• In Dennett’s example, an attitude is emergent from its own implementation, and
this suggests a way of saving Claim 1. While the machine doesn’t explicitly
represent “get your queen out early” it does explicitly represent some more
detailed rules of play – those that have “engineering reality” – for which the
strong form of Claim 1 is satisfied. However, this won’t solve the problem as
none of the principles by means of which a computational system operates need
be explicitly represented by formulae tokened in the device. In particular, the
program need not be explicitly represented; the point being that any function
computed by an algorithm can be hard-wired. So, what use is the computer
metaphor to RTM ?
• Worse, not all the rules of inference that a computational system runs on can be
represented just explicitly in the system; some have to be “realised in the
hardware” or the machine won’t run at all.
• Fodor agrees with this, but defers considering whether cases where programs are
hardwired defeat Claim 1 until after considering Case 2.

Case 2 : Mental Representations without Attitudes

• The feature of computers borrowed by RTM so far has been the mechanisation of
rationality, utilising a syntactically-driven machine to exploit the parallelisms
between the syntactic and semantic properties of symbols. Some versions of RTM
also borrow a theory of intelligence, wherein intelligent behaviour exploits a
cognitive architecture consisting of hierarchies of symbol processors. The top of
each hierarchy is a complex capacity while at the bottom is only the sort of
unintelligent operations characteristic of a Turing machine. The middle layers
form an explanatory reduction or analysis of intelligent capacities into dumb ones.
• Fodor gives an example of this sort of thing – tying one’s shoe-laces, in which a
homunculus follows instructions in a manual and in which various tasks are
broken out amongst gangs of wage slaves. At the top of the hierarchy are states
corresponding to propositional attitudes (knowing how to ties one’s shoes), but the
middle and bottom levels there is lots of symbol-processing that corresponds to
nothing that people, as opposed to nervous systems, do. While these lower states
satisfy Claim 1, in that they involve causally efficacious tokenings of mental
representations, they do not correspond to the tokening of an attitude as judged by
common sense. So, how can we modify Claim 1 to avoid counter-examples by
sub-personal information processing ?

Vindication Vindicated

16
I need to consider the import of this.

[email protected] Page 10 of 11
Jerry Fodor – Psychosemantics: Chapter 1 Notes

• Fodor says that these objections are not serious. What counts for the vindication
by RTM of belief/desire explanation is that in the core cases, rather than in all
cases, there should correspond a tokening of a mental representation for what
common sense counts as the tokening of an attitude (and vice versa). RTM has to
be able to say what it counts as a core case just as chemistry has to say what
counts as chemically pure water if it is to avoid “water is H2O” being
disconfirmed by water in a river.
• Fodor’s core cases are covered by the slogan “No intentional causation without
explicit representation”. That is, RTM is committed to the explicit representation
of all thoughts featuring as episodes in mental processes, otherwise it is false. This
follows from Claim 2, where mental processes are seen as causal sequences of
transformations of mental representations.
• This gets us out of the counter-examples. The rules that determine the
transformations of explicit tokenings of mental representations need not be
explicit but can emerge from the explicitly represented procedures of
implementation, or from the hardware. The “laws of thought” (modus ponens,
“get the queen out early”, …) may be explicitly represented, or not in these cases,
but the “data structures” (the contents of thoughts) have to be represented.
• So, in Dennett’s case “better get the queen out early” never constitutes an episode
in the mental life of the machine – ie. at the program level – it never consults this
rule in its operation. However, this is no problem for the core cases covered by
RTM. What is essential is that the actual or potential positions on the board be
explicitly represented, because the machine’s computations are defined over
these. These computations constitute the machine’s mental processes, so for a
representational theory of chess playing to be true of it they have to be causal
sequences of explicit representations. The relevance of the computer metaphor to
RTM arises from the connection between the constitution of mental processes as
computational and there being a language of thought17.
• Finally, what about the second counterexample – the sub-personal cases where
there are tokens of mental representation without tokens of attitudes ? What is
needed is that scientific psychology is ontologically committed to beliefs and
desires, not that the folk-psychology inventory of propositional attitudes be
exhaustive and co-extensive with that of RTM. What RTM needs to do is show
how intentional states could have causal powers. It is no problem for molecular
physics that it classifies glass as a liquid while common sense would not.
• In summary, common sense belief/desire explanation is vindicated if some good
theory of the mind is committed to entities like the attitudes which are both
semantically evaluable and etiologically involved. RTM is such a theory, so if it is
true, common sense is vindicated. RTM needs to provide an independently
confirmed empirical case for mental processes as causal sequences of
transformations of mental representations. This is the job of modern cognitive
psychology, which is almost totally dedicated to this end. A more strictly
philosophical problem turns on the semantic assumptions made by RTM, and this
is the subject of the rest of Fodor’s book.

17
Examine Fodor’s endnote on this point !

[email protected] Page 11 of 11