Julián Montaño 
Last weeks many newspapers echo the device designed by Carnegie Mellon University (Pittsburgh, US) researchers described in the last number of Science: Predicting Human Brain Activity Associated with the Meanings of Nouns (Science 30 May 2008: Vol. 320. no. 5880, pp. 1191 – 1195). Tom M. Mitchell and his team have designed a computational model that predicts brain activity produced when individuals think in different words. Until now we had mapped with resonances the different spatial patterns of neural activity with specific words. This computational model (well, a software) predicts these spatial patterns (the image obtained by resonance) for words for which we have not the image yet. If you are interested you can read the full article in www.sciencemag.org .
Nothing philosophical is relevant about the fact that you employ your department resources in a machine that advance you the sort of neural activity a brain shows when the individual listens to a particular word. Maybe could it be helpful for solving problems about language learning.
The relevant fact is when you try to draw too many conclusions not only from the possibility of predicting brain activity but from the fact in which it is supposed to be based upon this invention: that you can map brain activity through resonance images. Look at the title of the article: “…meaning of nouns”. And now look at some of the statements of these –on the other hand brilliant- researchers: when you think about an apple or hear the word apple a certain sort of brain activity is triggered, then the representation of that activity is a representation of the meaning of apple – “This suggests a theory of meaning based on brain function,” says Prof. Mitchell (quoted from the interview in http://esciencenews.com ). Brain function, the spatial pattern showed by the brain represents, draws, the meaning. This is to go too far.
Brain activity represents meaning no more than Bank activity represents money. Suppose we are studying a closed community where only one bank opens. They have this bank and their own money (in homage of brain sciences let us call it neuros instead of Euros). They do not practise economical transactions with any other and the bank does not operate with any other institution. It is easy for us to give a complete account of all the bank transactions: all the delivery of money (i.e. to express), all the stocking of money (i.e. the memory), all the new money put in circulation and financial products (i.e. to create), and all the exchanges (i.e. to translate). Does the bank transaction give us a complete account of money? No, since money is a social institution. For understanding money I have to understand what is to shop, to save money, to contract, to promised, the pattern under which money is valued, the individual property, the things that have no price in that community and even who is not allowed to carry money and who is allowed to, on top the bank activity, the assurance activity and so on. Money is a complex social institution that could not be localized in a place and described in a thin manner; it is a thick concept that needs a complex, non reductionist explanation.
All the same with meaning. I have mapped through electromagnetic resonance the spatial neural pattern of the word "apple". ¿Really does that represent the meaning of "apple"? Does this pattern represent the same meaning that "apple" has when it is hear by an individual who listen to his wife while she gives him the shopping list, when it is hear by an individual who listen to an account of the Original Sin and when it is hear by a child who listen the William Tell story?. Does "pomme", "maçã", "ringo" produce the same brain pattern? Context, bodily gestures and even the net of beliefs of both the speaker and the listener are part of Meaning. Meaning is a social institution: is the public use of a word, something that is irreducible to singles representations, something not to be itemized.
Is it possible that the same brain pattern that represents the meaning of "apple" is triggered by the word uttered by me when we are ordering the dessert in a restaurant, the word "apple" written in the label of a perfume, the image of an apple in the face of a man in a Magritte’s picture and the moving image of an apple in my online supermarket? Suppose that so, it is. Why these different cultural expressions end drawing the same brain pattern? "Well", a defender of meaning as brain representation could say, "they have a meaning in common therefore it is expected that they produce -though different human expressions as they are- the same brain pattern". OK, in this case Meaning is not the brain activity nor is represented by the same brain activity, since what explains that particular common brain activity is the meaning that they have in common. Meaning is other thing. "No", could answer again our scientist, "what I am trying to explain is that because they produce the same brain activity, they have the same meaning". Absolutely fine, but then, the fact that the same brain activity is produced by such (and potentially infinite in number) different expressions remain unexplained.
All these puzzles arise because the original malaise of AI (Artificial Intelligence) and many of the Brain Sciences projects is the mania of talking about language and meaning as something that happens in the head. It is a sort of reductionism, is like trying to reduce Monopoly game, with all the complexity it carries, to a mathematical formula. In the case of AI, the reductionism is very old, it comes from the idea that meaning and concepts are mental representations, pictures in the mind. Idea produced in the XVII c. and that jumped from philosophical Epistemology to Cognitive Sciences and Brain Sciences.