On friday I attended the first session of the CUNY Cognitive Science Speaker Series. The talk seemed to me to be based largely on this paper. I only have a few moments but I thought I would jot down the gist of the talk while it is fresh in my mind.
Godfrey-Smith wanted to take the ‘sender-receiver’ model of communication develop by David Lewis and apply it to debates about memory. On the Lewis model we have a sender that has access to a part of the world that the receiver does not, a sign that is passed between them, and a receiver that is able to take the sign and produce an action. Godfrey-Smith’s guiding idea is that when you have this kind of set up in the psychology of an organism and the signaling takes place over time, then you have memory.
One of the main points he wanted to make in the first half of his talk was that the idea of episodic memory as ‘constructive’ does not show that one of the main functions of episodic memory is to be truth preserving. He was aiming to oppose a group of scientist working on memory who hold what he called the ;future first’ hypothesis about episodic memory. Roughly speaking the idea is this. Our ability to imagine future events, and their outcomes is crucial for us and gives us an evolutionary advantage against those that cannot do it. What we have found out is that the neural areas that underlie our ability to do this are also largely the same ones involved in our ability to remember our past experiences. The ‘future first’ hypothesis is the idea that our ability to remember our own past experiences is simply a by-product of our ability to imagine future events and their outcomes. This is supposed to be further supported by the fact that episodic memory is thought of as ‘constructive’ in the sense that it is often wrong about the details of past experiences and tends to ‘construct’ memories along the most likely scenarios. Godfrey-Smith argued that if we think of memory in terms of the sender-receiver model then we should not immediately expect that the contrastive nature of episodic memory means that it is not truth-tracking. It is perfectly conceivable that the senders produce truth-tracking representations and that the receivers, who may be a bit smarter, ‘improvise’ from there. We can then go on and ask just how much deviation from the facts there is in the sender’s signal.
In the second half of his talk he went to discuss the controversy in cognitive science of whether there is any kind of reader in the brain. That is, is there anything in the brain which is akin to the ‘head’ in a turing machine. Something which interprets whatever message has been sent by the sender. He argued that the dominant view in the sciences is that there is no such reader. Godfrey-Smith went on to argue that there must be a reader, but that there is no sender. DNA, for instance, on this view is not an instance of information being sent. It is information, that happens to be able to be read (as a result of natural selection), but it has not been ‘written’ because, to put it crudely, nothing had the purpose of sending the message. In discussion he wanted to back away from talk of intention and purpose as ‘shorthand’ for the longer answer but I couldn’t make out what the longer answer was supposed to be.
One thought on “Peter Godfrey-Smith on Evolution And Memory”
[…] Peter Godfrey-Smith on Evolution And Memory […]