Patrick Stevens bio photo

Patrick Stevens

Former mathematics student at the University of Cambridge; now a software engineer.

Email Twitter Github Stackoverflow

I’ve been reading one of Daniel Dennett’s books, Consciousness Explained. Aside from the fact that the author has an incredible beard and is therefore correct on all matters, he can also write a very cogent book. In Consciousness Explained, Dennett outlines what he calls the Multiple Drafts approach to explaining consciousness; this blog post is my attempt to summarise that view in a couple of short analogies.

Dennett starts off by providing evidence that our time-perception is somewhat malleable: we can interpret two dots of different colours (appearing separated by a short distance in time and space) as a single moving dot that changes colour abruptly at some point. The key puzzle here is that we perceive the colour to have changed before seeing the second coloured dot. Dennett then outlines what seem to be the two mainstream points of view on how this happens.

  • The Orwellian view: that at the time of perception, we saw exactly what happened, and then we edit this after the fact to reflect a more logical sequence of events (à la Minitrue);
  • The Stalinist view: that the information is edited before even making its way into the consciousness.

Dennett points out that both of these options implicitly assert the existence of a “Cartesian Theatre” - a place where consciousness is experienced as information is gathered. In particular, the Stalinist view requires consciousness to be experienced after sufficient time has passed for some decisions to be made. By the way, in arguing against this supposition, Dennett doesn’t mention that there is precedent for this kind of behaviour in the reflex action, which we explicitly only realise we have made after it has happened; but it’s a minor point, since there are sound physiological reasons for why the reflex action doesn’t come under conscious control (the signal for action never actually enters the brain, but is headed off at the brain stem). He then gives a third possible view - the Multiple Drafts model. In each of the next two analogies, I will liken the consciousness to a general in war, making decisions based on reports from the battlefield. In fact, Dennett argues that since the Cartesian Theatre does not exist (that is, consciousness isn’t something that is recorded and played back to some internal watcher), this type of analogy is deeply flawed, and the third analogy will contain an appropriate adjustment.
Central to the analogies are two reports in particular:

  1. “At location X at time 15:00:00, M happened”, analogous to the report-to-the-consciousness “My hand tells me that I drew near to a source of intense heat at time ___”;
  2. “At location X at time 15:00:02, N happened”, analogous to the report-to-the-consciousness “My eyes tell me that I touched the hot plate at time ___”.

We consider the case that report 2 arrives before report 1 (even though report 2 describes events which occurred later than report 1) - this is quite conceivable given the distance that messages must travel in the nervous system. (Please ignore the fact that this particular effect is probably going to work in reverse for this particular example, the eyes being closer to the brain than the hand - and assume that every decision is made in the brain, so that reflexes don’t happen. It’s harder than you might think to come up with something sufficiently urgent that isn’t made as a reflex!)

The Stalinist analogy

In this version of events, the reports come in from the battlefield, and flow through the general’s underlings. The underlings see that the reports are in the wrong order, and switch them round so that they are in the right order, before presenting them to the general in the order {2,1} to consider; they also decide that there is a missing piece of information [corresponding to the “change-in-colour-of-dot” situation, but that doesn’t fit with this analogy] between reports 2 and 1, so they insert it. The general acts on the augmented reports, and they are then sent off to be filed away for future reference.

The Orwellian analogy

In this version of events, the reports come in from the battlefield, but the underlings don’t correct the order of the reports, so the general sees {1,2}. The general acts on the reports once they’ve both been received, noticing that some information seems to be missing and adding it in, and sends them off to be filed. The archivist sees that they are in the wrong order, and switches them round just before filing them.

The Multiple Drafts analogy

In this version of events, the reports come in one by one from the battlefield, but there is no general - just a room full of underlings. The first report (which records a later event) comes in, and the underlings all update their states-of-mind accordingly. Then the second report (which records an earlier event) comes in; the underlings nearest the door update soonest, and the report makes its way around the room from underling to underling. The underlings act on the reports (Multiple Drafts doesn’t address how this happens - for the purposes of this analogy, let it be by everyone shouting at once, and the majority view prevails). As time goes on, more reports flood in, but eventually every underling has received reports 1 and 2 (this may happen before or after the action based on those reports is taken), and the archivist-underling files what ey thinks happened.

Under Multiple Drafts, then, there is never a “point at which information enters the consciousness”, but rather a “time interval in which information is making its way around the consciousness”. The name of the model comes from an analogy to writing a summary of the events - starting from report 1, a summary is written; then report 2 is added, it progresses around the consciousness, and wherever it arrives, the summary is updated to reflect the new information. Thus there are multiple drafts of the summary at once. When the information is fully incorporated (that is, consensus has been reached on what the summary should contain), the consciousness is free to store the consensus draft in memory for future reference. Note that this could happen some time after the events described in the summary - Dennett is careful to separate “what happened” from “how the consciousness stores what happened”.

The reason Multiple Drafts is so attractive is that there is no experimental way to differentiate between Orwellian and Stalinist. Either way, the subject of an experiment will report the same thing, so it is strange to draw a distinction between these two possible methods. Having noticed that Orwellian and Stalinist are indistinguishable, the natural question is “why do we think they are different?” - and it turns out that the only real reason is that we think there is a centre of consciousness, through which the information must flow. Only under that interpretation is there a difference between amending-before-consideration and amending-after-consideration. So we relax the assumption of a centre of consciousness, and we end up with a “smear” of time during which information is incorporated, rather than an absolute time of perception. (This is borne out by experiment, by the way - we are very flexible when it comes to simultaneous perception.) This idea makes sense - we don’t perceive space absolutely, and can happily work with receiving information about space at smeared-out times, adding more information to the model as we find out more. I nudge the table-leg with my foot, someone reacts, I am swinging my foot to kick it again, but just when I can no longer stop the kick I realise that it was in fact a human leg, the other person glares at me - my perception of the layout of space below the table has developed as new information came in, but out of sync with the information itself (the information bunched up and all came along at once). There is no particular reason why our time-perception should be any different.

The book is an excellent one, very coherently written - this blog post doesn’t really do it justice (although that’s the point of this blog - to get me practised at writing). As of this writing, I am only half-way through the book, but it is shaping up well.