Wednesday, September 30, 2009

Distinguishing between apparent and real time in reporting Libet-related research

Here's a distinction that many writers seem insufficiently sensitive to in reporting research by and about Libet and the questioned efficacy of conscious will: It's the distinction between when something seems to have happened and when something really happened. This is a distinction we might label as that between subjective and objective time or apparent and real time.

It seems easy enough to acknowledge such a distinction and find clear cases of its applicability. Unbeknownst to George, his watch is 5-mins fast. So, when some event actually happened may be a quite different time (a difference of 5 mins) than when it seemed to George to have happened. This is so simple, that young children can keep them apart. So whence the alleged difficulty?

Things start to get a bit confusing when we realize that mental events we can call seemings can have real objective times of occurrence too. Especially confusing is when the seemings concern times of occurrence. Ordinary language is ill-equipped to keep the real time/apparent time distinction clear in describing such situations. (My spell-checker is hating on "seemings" right now.) Suppose George's seeming objectively occurred at 12:50, but it seemed to George that the mental event in question occurred at 12:55. Note how totally ambiguous a statement like the following can seem once such a distinction is in mind:

"It seemed to George that he made a conscious decision at 12:55 to flick his wrist"

Is 12:55 here the objective time of the seeming or the subjective time of occurrence according to George?

Here's some recent blogging on Libet that seems to me to make the mistake I'm talking about.

From "I didn’t do it, my brain did" @The Philosopher's Eye

In 1983 Libet showed that before subjects announced their decision to perform an action (and hence, or so Libet assumed, before deciding to perform an action) their motor cortex was already preparing the way for the act in question.

It seems that the safe thing to say of what Libet showed is that the apparent (according to the subject) time of the conscious descision was later than the objective time of the (perhaps conscious) decision.

From "
Libet was wrong...?" @Conscious Entitites

...Benjamin Libet’s remarkable experiments, which seemed to show that decisions to move were really made half a second before we were aware of having decided.

Here the crucial ambiguity is with "half a second before we were aware of having decided." Is that 1/2second subjective? If so, that's a modest, though unexciting, description of what Libet showed. Is it instead objective? That would be very exciting indeed, but it's a great mystery how Libet, or anyone, could possibly show such a thing.

Anyway, the line I'm recommending here is what I take to be be one of Dan Dennett's key insights in his various discussions of Libet, so I don't pretend any originality here.

5 comments:

  1. I think that timing is key, but not the way you refer to the timing. Anyone who has played a sport like baseball knows that when you hit the ball the timing is a key skill, and that the skill of knowing when to move requires practice until the brain knows how to manage the timing without a lot of conscious processing, so the the conscious brain can concentrate on tactics, communication and strategy.

    So why should we be surprised that the conscious mind delegates timing of similar events? This does not mean that we do now decide to do, just that the automatic brain is better at deciding when to to what we already have decided to do (within the span of working memory, that is).

    ReplyDelete
  2. While it is important to keep the time of representations distinct from the representation of times, I'm not sure that things are all that muddled in standard interpretations of Libet.

    Stepping back, we can ask when someone experienced some percept (e.g., a Where's Waldo task when they point at his face when they see it in the picture). While there is some slop due to reaction time issues, let's say they saw Waldo at the time at which they pointed. We could find out how much longer it takes for someone to experience Waldo when the picture is larger and more complicated. Again, assuming the same reaction time slop in the two cases, we could get a delta t and get a fairly precise answer.

    Similarly, we can have people indicate when they experience a sensation, for instance if we inject currents of different magnitudes (Penfield style) into the somatosensory cortex and they must tell us when they experience it. In this case we might have them look at a digital clock and tell us the time at which they experienced it. Again, there will be some slop, but that's biology. We would find that we can give a small amplitude current for a long time, or a large amplitude current for a shorter time, and I'd take that as evidence that different stimuli take more time (objectively speaking) to evoke a conscious percept. Since reaction time slop is probably similar in the two cases, we could again get a delta t measure of the objective difference in time of onset of the experiences.

    I'm not sure what isn't clear here. Whether they perceive the sensation as happening "now" or not, we can measure the effects of the representational vehicle doing the representing.

    If you buy all this, what is wrong with saying that Libet showed that the (objective) time between the readiness potential and the time at which subjects observed "Waldo" (i.e., their intention to act) was 500 ms?

    Consciousness is a weird thing, as even my experience imagining something happening in the past isn't experienced as happening in the past, it is a present experience of a memory of something in the past. It seems one thing that makes conscious (especially perceptual) contents special is that they represent things as happening now, so the vehicle and content seem to align fairly nicely, with some slop. This seems an empirical question.

    Dennett makes a great deal of the variation and slop in timing of perceptions and perceptions of timing, for instance when you really push things down to the millisecond time scales. I just chalk it up to variability and imprecision in our measurements and theories right now, not the rather premature conclusion Dennett reaches that there is no fact of the matter about the timing of experiential contents.

    I experienced the Patriots' tragic loss in February 2008 before the tragic loss of Brady in September 2008. Both objectively (in terms of timing of the representations doing the experiencing), and subjectively (in terms of how I represent those two things) occurred in a particular order, and the objective time can fairly cleanly be connected to the subjective time using methods like those above (the Where's Waldo methods).

    The topic is complicated, and with the property of time things do get very strange because we always seem to experience what is happening "now" (as opposed to spatial experience where we see things as over here, over there, all over the place). That is, when it comes to time, there may be a natural alignment between the representer, and the represented, one that you don't find with any other properties.

    In other words, three cheers for the Cartesian Theatre, where the timing of objective neuronal events maps fairly nicely (with some variability of course) onto the movie experienced in the brain.

    ReplyDelete
  3. Let's assume George is looking at a clock and correctly perceives the timing of some event, say, a light flashing at t. Let us also assume that the perception is a conscious perception and George accurately reports the content of his perception.

    The following all seems pretty safe: The conscious percept objectively happened after t. The accurate report objectively happened after t. The intention to accurately report happened objectively before the the report itself.

    It remains unclear to me how you can get much more and with more precision than that. So it remains unclear to me how the Cartesian theater doesn't suck.

    ReplyDelete
  4. There is certainly nothing incoherent or violated by data in imagining a verified model of a consciousness box that tells us when X happens, subject is experiencing an intention to act thusly. We see that X happens 500 ms after the readiness potential.

    Because we don't have enough data or a good enough model to say something like that, I agree it becomes hard to interpret Libet. That is, in these tough cases where we are down to hundreds of milliseconds, relying on people's reports and comparisons of different conscious contents, things do get tricky. I realize that was your main point, and agree with it.

    However, I am an fallabilist optimist. We can use common knowledge to put tighter bounds on our estimates of what is happening. For instance, what is the delay between the clock being at point T and subject seeing it at point T? We can put reasonable upper and lower bounds on this (e.g., much of the reaction time delay is from transduction at the retina, which puts a lower bound on the time).

    If, once we do reasonable calculations, the delay between readiness potential and consciousness of intention (as measured by the clock comparison (really a blip on an oscilloscope in his experiments)) is large enough, then we'd have a more conclusive result.

    While I haven't done the calculations, my hunch is that 500 ms is long enough to draw some conclusions. Wikipedia just told me that human reaction time to a visual stimulus is under 250 ms, so Libet's interpretation seems reasonable if not failproof, because the observed 500 ms delay is quite a bit longer.

    Of course, it isn't knock-down super transparent self-interpreting result, but quite suggestive, and I believe subsequent work with MRI has seen much longer times that make things even more clean (paper here).

    Whether the Cartesian theatre exists is an open empirical issue. Dennett jumped the gun, and philosophers for some reason jumped right with him (e.g., the slogan 'there is no place where it all comes together' is taken as nearly axiomatic, as I railed against here).

    So, while things are a bit messy, and I agree that the statement 'The readiness potential occurred 500 ms before the conscious intention to act' is an over- interpretation, it isn't all that bad as it likely gets things qualitatively right.

    ReplyDelete
  5. "The measurements of brain activity confirm the obvious logical point that you cannot know you have made a decision until you have made it!" (Conscious free will and empiricism). So any experiment that did not demonstrate a readiness potential would be false.

    The debate about "conscious free will" is actually a debate about the nature of time. I can never decide to do anything in an instant so my decisions must all have been in the past. I cannot be conscious in an instant so how do I know I am currently conscious?.

    ReplyDelete