Let us go back to Fodor’s claim (3). Suppose we think that the causal transitions holding between sentences in the language of thought are essentially syntactic, holding purely in virtue of the formal properties of the relevant symbols irrespective of what those symbols might refer to. Then we need to ask the following question.
Why do the syntactic relations between sentences in the language of thought map onto the semantic relations holding between the contents of those sentences?
If we take seriously the idea that the language of thought is a formal system, then this question has a perfectly straightforward answer. Syntactic transitions between sentences in the language of thought track semantic transitions between the contents of those sentences for precisely the same reason that syntax tracks semantics in any properly designed formal system.
Fodor can (and does) appeal to well-known results in meta-logic (the study of the expres- sive capacities and formal structure of logical systems) establishing a significant degree of correspondence between syntactic derivability and semantic validity. So, for example, it is known that the first-order predicate calculus is sound and complete. That is to say, in every well-formed proof in the first-order predicate calculus the conclusion really is a logical consequence of the premises (soundness) and, conversely, for every argument in which the conclusion follows logically from the premises and both conclusion and premises are formulable in the first-order predicate calculus there is a well-formed proof (completeness).
Put in the terms we have been employing, the combination of soundness and com- pleteness has the following important consequences. If a series of legitimate and formally definable inferential transitions lead from formula A to a second formula B, then one can be sure that A cannot be true without B being true– and, conversely, if A entails B in a semantic sense then one can be sure that there will be a series of formally definable inferential transitions leading from A to B.
Let’s look at an example of how this is supposed to work. Suppose that we have two complex symbols. Each of these symbols is a sentence in the language of thought. Each has a particular syntactic shape. Let us say that these are Ga and Fa respectively. These syntactic shapes have meanings – and the particular meanings that they can have are a function of their shape. We know that“F–” and “G–” are symbols for predicates. Let us say that“F–” means “– is tall” and “G–” means “– has red hair.” We also know that “a” is a name symbol. Let us say that “a” names Georgina. The meaning of “Ga” is that Georgina is tall, while the meaning of“Fa” is that Georgina has red hair. We can look now at how a very simple piece of thinking might be analyzed by the language of thought hypothesis.
In the table we see how two physical symbols:“Ga” and “Fa” can be transformed in two inferential steps into the more complex physical symbol“∃x (Gx & Fx).” The rules that achieve this transformation are purely syntactic, in the sense that they are rules for manipulating symbol structures. But when we look at the relation between the mean- ings of“Fa” and “Ga,” on the one hand, and the meaning of “∃x (Fx & Gx)” on the other, we see that those purely syntactic transformations preserve the logical relations between the propositions that the symbols stand for. If it is true that Georgina is tall and that Georgina has red hair, then it is certainly true that at least one person is tall and has red hair.
To draw the threads together, then, beliefs and desires are realized by language-like physical structures (sentences in the language of thought) and practical reasoning and other forms of thinking are ultimately to be understood in terms of causal interactions between those structures. These causal interactions are sensitive only to the formal, syntactic properties of the physical structures. Yet, because the language of thought is a formal language with analogs of the formal properties of soundness and completeness, these purely syntactic transitions respect the semantic relations between the contents of the relevant beliefs and desires. This is how (Fodor claims) causation by content takes place in a purely physical system such as the human brain. And so, he argues,
S Y M B O L S T R A N S F O R M A T I O N R U L E M E A N I N G
1. Ga 1. Georgina is tall
2. Fa 2. Georgina has red hair
3. (Fa & Ga) If complex symbols “S” and “T” appear on earlier lines, then write “(S & T)”
3. Georgina is tall and has red hair 4.∃x (Fx & Gx) If on an earlier line there is a complex
symbol containing a name symbol, then replace the name symbol by “x” and write“∃x –” in front of the complex symbol
4. At least one person is tall and has red hair
commonsense psychological explanation is vindicated by thinking of the mind as a computer processing sentences in the language of thought.
The line of reasoning that leads to the language of thought hypothesis is fairly complicated. To make it easier to keep track of the different steps I have represented them diagrammatically inFigure 6.3
Exercise 6.7 Use the flow chart inFigure 6.3to explain Fodor’s argument in your own words.
Successful practices of belief–desire explanation
Intentional realism
Problem of causation by content
Distinction between formal properties and semantic
properties
Language of thought (LOT) hypothesis
Syntactic level
Sentences in the (LOT) interacting in virtue of their formal properties Logical deducibility
Semantic level
Propositions that stand in logical relations
to each other Logical consequence
Figure 6.3 The structure of Fodor’s argument for the language of thought hypothesis.
6.3
The Chinese room argument
The physical symbol system hypothesis holds that we have intelligent behavior when (and only when) we have systems that manipulate symbols according to rules. The language of thought hypothesis is a particular way of applying this model of intelligent behavior. It offers a specific proposal for how to understand the symbols. The symbols are sentences in an internal language of thought. The language of thought hypothesis also tells us what the rules are going to be like and how they will end up producing intelligent behavior. These rules are fundamentally syntactic in form, transforming the physical symbols in ways that depend solely on their physical/formal characteristics. These transformations will produce intelligent behavior because syntactic transform- ations of the physical symbols mimic semantic relations between the propositions that give meaning to the physical symbols.
We need now to stand back from the details of the language of thought hypothesis to consider a fundamental objection to the very idea of the physical symbol system hypothesis. This objection comes from the philosopher John Searle, who is convinced that no machine built according to the physical symbol system hypothesis could pos- sibly be capable of intelligent behavior. He tries to show that the physical symbol system hypothesis is misconceived through a thought experiment. Thought experiments are a very standard way of arguing in philosophy. Thought experiments are intended to test our intuitions about concepts and ideas. They do this by imagining scenarios that are far-fetched, but not impossible, and then exploring what we think about them.
The basic idea that Searle takes issue with is the idea that manipulating symbols is sufficient for intelligent behavior– even when the manipulation produces exactly the right outputs. What he tries to do is describe a situation in which symbols are correctly manipulated, but where there seems to be no genuine understanding and no genuine intelligence.
Searle asks us to imagine a person in what he calls a Chinese room. The person receives pieces of paper through one window and passes out pieces of paper through another window. The pieces of paper have symbols in Chinese written on them. The Chinese room, in essence, is an input–output system, with symbols as inputs and outputs. The way the input–ouput system works is determined by a huge instruction manual that tells the person in the room which pieces of paper to pass out depending on which pieces of paper she receives. The instruction manual is essentially just a way of pairing input symbols with output symbols. It is not written in Chinese and can be understood and followed by someone who knows no Chinese. All that the person needs to be able to do is to identify Chinese symbols in some sort of syntactic way– according to their shape, for example. This is enough for them to be able to find the right output for each input – where the right output is taken to be the output dictated by the instruction manual.
The Chinese room is admittedly a little far-fetched, but it does seem to be perfectly possible. Now, Searle continues, imagine two further things. Imagine, first, that the
instruction manual has been written in such a way that the inputs are all questions in Chinese and the outputs are all appropriate answers to those questions. To all intents and purposes, therefore, the Chinese room is answering questions in Chinese. Now imagine that the person in the room does not in fact know any Chinese. All he is doing is following the instructions in the instruction manual (which is written in English). The situation is illustrated inFigure 6.4. What the Chinese room shows, according to Searle, is
[Whoever or whatever is in that room is an intelligent Chinese speaker!] Take a squiggle
squiggle sign from tray number 1 and put it next to a squoggle-squoggle sign from basket number 2. “I’m just manipulating squiggles and squoggles but I don’t really understand what they mean. This rule book written in
English tells me what to do. I get the squiggle squiggle from here,
look at the book, and then move the squoggle squoggle over there.