Aug 042006

Earlier this week I spoke to Andreas Weigend’s statistics class at Stanford University. In such a setting, one of the points I wanted to make is that attention is not absolutely quantifiable and never will be.

Is my full attention numerically the same as yours? How could we find this out? Is my full attention at one time, equal to that at another, or if less, in what proportion? If I am aware that I am not paying full attention to you right now, can I possibly say the exact percentage of attention I am paying? I do not think there is any way to do any of this with any significant degree of precision at all.

What one can do, which is quite common, is to make an ordinal comparison of relative audience attention. That is the basis of bestseller lists, Nielssen ratings, comparisons of movie grosses, measuring the number of visits to sites on the Internet, etc. (They are ordinal, that is they tell us the relative size of audiences without in any way claiming to tell us the intensity of attention of the average audience member. The bestseller list one week simply tells us the order in which books are sold, not even the absolute number of books in the top spot. But even that knowledge, which might be available, would not tell us even relatively how attentively the books are read, or even if they are read at all.)

What about more direct physiological measures? It may be possible to measure skin potentials and the like which say something about mental arousal. Let’s suppose that many thousands of people are hooked up to such devices, with time-correlated output, as they are watching, say, the Academy Awards. Then spikes in arousal for the whole group might tell us about relative attention paid to different aspects of the program, but how that would even translate into proportionate intensity of attention would remain unclear. Is the arousal due to what happens that second, or is it possibly a delayed pondering of what just happened? Is it only due to attention per se, or might other variables, such as emotional content affect the results?

In my Stanford presentation, I mentioned mirror neurons and their role in attention as briefly discussed in my June First Monday article. A student asked whether, if neurons are involved in attention, somehow counting their activity might lead to the quantifiability of attention. I was unsure in principle, but upon reflection, I don’t think this would ever work.

If you put someone in a magnetic resonance scanner, you can see how much different parts of her brain are active at different moments. But can you even say with any certainty where the attention is going? Some of it — most, perhaps — is actually going to the experimenters who got her to agree to get into the contraption. Some of it may be focused on her own desires to get out. It is always possible that whatever she is supposed to be paying attention to evokes memories that focus most of her attention somewhere else. Unless you could map her brain so completely as to correlate every synapes’s firing with some definite thought, conscious or unconscious, you could never answer such questions. At a minimum, to do that, you would have to have her in the scanner her entire life, and everything she took in would have to be carefully attended to by several researchers acting independently, for any reliability. Even with a highly miniaturized and much more convenient form of scanner than anyone now anticipates, this cannot become a general procedure, quite apart from its ethical dubiousness.

Human brains possess something like twenty billion neo-cortical neurons, and perhaps one hundred billion cortical neurons, though more than the cortex may be involved in attention. Each neuron can have thousands of synapses connecting it with others. On this basis too, the idea that being able to know numerically what is going on in someone else’s brain seems highly improbable.

And that would be the sort of thing one would want to know if the attention economy were to attain the apparent quantitative aspects of the money economy. Presumably, it would be theoretically possible to take a “snapshot” of someone’s monetary worth at a particular time, by noting the value on the market at the moment of stocks, money, real estate, and other possessions. But no such estimate with similar precision would be available for ascertaining just how many people would pay you how much attention now were you to seek it.

This does not mean that you have no attention wealth, or more broadly that the attention economy cannot function, but just that it must function differently. It has no banks, no loan guarantees, no futures markets, etc.

  3 Responses to “Can You Count Attention Units?”

  1. Security personnel at airports are facing arguably analogous problem: how they could ensure that screeners pay enough attention to pictures at their X-Ray screens. They came up with an ingenious solution – the software randomly project pictures of dangerous objects, like guns, or knives on top of pictures of moving luggage. If the screener has not noticed them it means, he does not pay enough attention. This is not exactly “measuring” attention, but at least ensuring certain level of it.

  2. michael, as you know i respectfully disagree. in his investogations #71, wittgenstein asks “Is an indistinct photograph a picture of a person at all? Is it even always an advantage to replace an indistinct picture by a sharp one? Isn’t the indistinct one often exactly what we need? ” while the total flow of information may ever be entirely quantifiable, that does not mean that representative proxies of it (through various forms of data analysis) cannot be traced and provide insights. these insights would be specifically tied to the “counting” of attention units (aka attentrons). and so even if this counting was indistinct, it doesnt mean that it would be unusable.

  3. Thanks for comments. I don’t think there is a real disagreement. With sufficient fuzziness, one can offer estimates of attention. When two estimates diverge sufficiently, and seem fairly robust we can learn something. For instance, Christina Aguillera gets lots more attention than the woman who sings once a month in the local club.

    Or take the method used to check on airport security personnel. (By coincidence I came up with the same scheme right after 9-11, but I thought it was too obvious as well as too much trouble to do anything about.) Actually such spot checks on attention are common in a variety of circumstances. Still, they don’t precisely measure attention at all. They measure some combination of attention and visusal observing ability, for instance, and of course it is the combined level that matters in this case. The same security officer who passes the test might be far more attentive at say a football game or a rock concert. And unless the sample objects are infinitely varied, they will not necessarily demonstrate that the guards who see them are the same ones who might spot the next shoe bomber or some other unexpected anomaly. It can become rather like driving; if you are familiar with a road, you may not be very attentive most of the time. A guard might become familiar with the fake test pictures.

    A psychoanalyst or a music critic would probably have to be much more attentive, but we can’t easily measure their performance by a test. We can only gauge it by paying complex and lengthy attention to either of them ourselves.

    All of this suggests that we cannot count attention the way we may count the silverware or barrels of oil or medical doses. In many instances this might make no difference. I still maintain that overall it amounts to an important distinction.

 Leave a Reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>