Quando Libet

New light on Libet’s challenge to free will; this interesting BQO piece by Ari N Schulman focuses on a talk by Patrick Haggard.

Libet’s research has been much discussed, here as well as elsewhere. He asked subjects to move their hand at a time of their choosing, while measuring neural activity in their brain. He found that the occurrence of a ‘Readiness Potential’ or RP (something identified by earlier researchers) always preceded the hand movement. But it also preceded the time of the decision, as reported by subjects. So it seemed the decision was made and clearly registered in brain activity as an RP before the subjects’ conscious thought processes had anything to do with it. The research, often reproduced and confirmed, seemed to provide a solid scientific demonstration that our feeling of having free conscious control over our own behaviour is a delusion.

However, recent research by Aaron Schurger shows that we need to re-evaluate the RP. In the past it has been seen as the particular precursor of intentional action; in fact it seems to be simply a peak in the continuing ebb and flow of neural activity. Peaks like this occur all the time, and may well be the precursors of various neural events, not just deliberate action. It’s true that action requires a peak of activity like this, but it’s far from true that all such peaks lead to action, or that the decision to act occurred when the peak emerged. If we begin with an action and look back, we’ll always find an RP, but not all RPs are connected with actions. It seems to me a bit like a surfer, who has to wait for a wave before leaping on the board; but let’s suppose there are plenty of good waves and the surfer is certainly not deprived of his ability to decide when to go.

This account dispels the impression that there is a fatal difficulty here for free will (of course there are plenty of other arguments on that topic); I think it also sits rather nicely with Libet’s own finding that we have ‘free won’t’ – ie that even after an RP has been detected, subjects can still veto the action.

Haggard, who has done extensive work in this area, accepts that RPs need another look; but he contends that we can find more reliable precursors of action. His own research analysed neural activity and found significantly lowered variability before actions, rather as though the disorganised neural activity of the brain pulled together just before an action was initiated.

Haggard’s experiments were designed to address another common criticism of Libet’s experiments, namely the artificiality of the decision involved. Being told to make your hand for no reason at a moment of your choosing is very unlike most of the decisions we make. In particular, it seems random, whereas it is argued that proper free will takes account of the pros and cons. Haggard asked subjects to perform a series of simple button-pushing tasks; the next task might follow quickly, or after a delay which could be several minutes long. Subjects could skip to the next task if they found the wait tedious, but that would reduce the cash rewards they got for performing the tasks. This weighing of boredom against profit is much more like a real decision.

Haggard persuasively claims that the essence of Libet’s results is upheld and refreshed by his results, so in principle we are back where we started. Does this mean there’s no free will? Schulman thinks not, because on certain reasonable and well-established conceptions of free will it can ‘work in concert with decisional impulses’, and need not be threatened by Haggard’s success in measuring those impulses.

For myself, I stick with a point mentioned by Schurger; making a decision and becoming aware of the decision are two distinct events, and it is not really surprising or threatening that the awareness comes a short time after the actual decision. It’s safe to predict that we haven’t heard the last of the topic, however.

Beyond Libet

Picture: dials. Libet’s famous experiments are among the most interesting and challenging in neuroscience; now they’ve been taken further. A paper by Fried, Mukamel and Kreiman in Neuron (with a very useful overview by Patrick Haggard) reports on experiments using a number of epilepsy patients where it was ethically possible to implant electrodes and hence to read off the activity of individual neurons, giving a vastly more precise picture than anything achievable by other means. In other respects the experiments broadly followed the design of Libet’s own, using a similar clock-face approach to measure the time when subjects felt they decided to press a button. Libet discovered that a Readiness Potential (RP) could be detected as much as half a second before the subject was conscious of deciding to move; the new experiments show that data from a population of 250 neurons in the SMA (the Supplementary Motor Area) were sufficient to predict the subject’s decision 700 ms in advance of the subject’s own awareness, with 80% accuracy.

The more detailed picture which these experiments provide helps clarify some points about the relationship between pre-SMA and SMA proper, and suggest that the sense of decision reported by subjects is actually the point at which a growing decision starts to be converted into action, rather than the beginning of the decision-forming process, which stretches back further. This may help to explain the results from fMRI studies which have found the precursors of a decision much earlier than 500 ms beforehand. There are also indications that a lot of the activity in these areas might be more concerned with suppressing possible actions than initiating them – a finding which harmonises nicely with Libet’s own idea of ‘free won’t’ – that we might not be able to control the formation of impulses to act, but could still suppress them when we wanted.

For us, though, the main point of the experiments is that they appear to provide a strong vindication of Libet and make it clear that we have to engage with his finding that our decisions are made well before we think we’re making them.

What are we to make of it all then? I’m inclined to think that the easiest and most acceptable way of interpreting the results is to note that making a decision and being aware of having made a decision are two different things (and being able to report the fact may be yet a third). On this view we first make up our minds; then the process of becoming aware of having done so naturally takes some neural processing of its own, and hence arrives a few hundred milliseconds later.

That would be fine, except that we strongly feel that our decisions flow from the conscious process, that the feelings we are aware of, and could articulate aloud if we chose, are actually decisive. Suppose I am deciding which house to buy: house A involves a longer commute while house B is in a less attractive area. Surely I would go through something like an internal argument or assessment, totting up the pros and cons, and it is this forensic process in internal consciousness which causally determines what I do? Otherwise why do I spend any time thinking about it at all – surely it’s the internal discussion that takes time?

Well, there is another way to read the process: perhaps I hold the two possibilities in mind in turn: perhaps I imagine myself on the long daily journey or staring at the unlovely factory wall. Which makes me feel worse? Eventually I get a sense of where I would be happiest, perhaps with a feeling of settling one alternative and so of what I intend to do. On this view the explicitly conscious part of my mind is merely displaying options and waiting for some other, feeling part to send back its implicit message. The talky, explicit part of consciousness isn’t really making the decision at all, though it (or should I say ‘I’?) takes responsibility for it and is happy to offer explanations.

Perhaps there are both processes in involved in different decisions to different degrees. Some purely rational decisions may indeed happen in the explicit part of the mind, but in others – and Libet’s examples would be in this category – things have to feel right. The talky part of me may choose to hold up particular options and may try to nudge things one way or another, but it waits for the silent part to plump.

Is that plausible? I’m not sure. The willingness of the talky part to take responsibility for actions it didn’t decide on and even to confect and confabulate spurious rationales, is very well established (albeit typically in cases with brain lesions), but introspectively I don’t like the idea of two agents being at work I’d prefer it to be one agent using two approaches or two sets of tools – but I’m not sure that does the job of accounting for the delay which was the problem in the first place…

(Thanks to Dale Roberts!)