Bergson, Time and Free Will, Chapter 1, §47 "Delboeuf's Results Seem More Plausible, but, in the End, All Psychophysics Revolves in a Vicious Circle"

by Corry Shores
[Search Blog Here. Index-tags are found on the bottom of the left column.]

[Central Entry Directory]
[Bergson, Entry Directory]
[Bergson Time and Free Will, Entry Directory]

[The following is summary; my commentary is in brackets.]

Bergson, Time and Free Will

Chapter I, "The Intensity of Psychic States"

Part XIV: "Psychophysics"

§47 "Delboeuf's Results Seem More Plausible, but, in the End, All Psychophysics Revolves in a Vicious Circle"

There is a psychophysical method that

1) measures sensation by standardizing sensation-units, and

2) finds the formula that determines quantitatively their correspondence to stimulus-variations.

Bergson has extensively examined and critiqued this method. He does not believe that it has advanced psychophysics. (67bc) [For more on this method, see this entry on Fechner's Law and the entries for §41; §42; §43; and §44.]

We also examined Delboeuf's ring experiment, which seems to confirm Fechner's theory. In his experiment, we simultaneously sense three light sources arranged in an order of brightness. They were adjusted until we detected a common difference: the 1st is as different to the 2nd as much as the 2nd is as different to the 3rd. Hence this seems to demonstrate that we may standardize sensation-quantities. This way we could measure any sensation's intensity.

So we are consider a change in sensation from S to S'. Fechner wanted to treat all sensation-increases as belonging along a continuum. However, we do not experience each increase continuously. Rather, we feel them proceed one-by-one as discrete changes or leaps to the next sensation. Thus Fechner's mistake is that he believed there was a continuous interval between S and S'. Instead, there is just a passing from one sensation to the next. There is no arithmetical quantity to the distance between S and S'.

But Delboeuf's experiment shows that we can have a array of simultaneous stimuli that differ one-by-one according to a stardard difference of sensation. It seems plausible then to consider the difference between S and S' as a quantity. Because both these simultaneous sensations are perceptions of light, they belong to the same genus. Also, we have many past experiences where we have perceived the continuous change of luminocity as it passes through these standardly different stages. Hence we might confuse the contrast between standardly different simultaneous sensations with ones that change continuously from one level to the next. In other words, we might regard them as arithmetically different. Also, even when stimuli are increased steadily, we might notice certain determinate jumps in sensation. So for this reason too we might think we can quantify sensations numerically.

To sum up, the contrast will appear to us as a difference, the stimulus as a quantity, the sudden jump as an element of equality: combining these three factors, we shall reach the idea of equal quantitative differences. (68d)

So we noted that we have experiences of light increasing gradually. This causes us to undergo a series of different sensation states. For Bergson, they are each qualitatively different, although there is a determinate number. But because a series of discrete states corresponds to a continuous stimulus, we mistakenly think that we may quantify the sensation for different levels of stimuli. For, we might transpose our experiencing a number of qualitatively different sensation states onto a magnitude presented in only one such state. So even in Delboeuf's experiment, we first have certain sensations, which then serve to determine the way that we measure subsequent sensations.

Likewise with Fechner's method. Experiments show that we have a certain fixed ratio between the starting-stimulus-quantity and the least amount we need to add to it in order to cause a change in sensation. Then we use these empirically determined ratios to measure other sensations.

In both cases, we are trying to find a standard of sensation-change that correlates to a standard of stimulus-change.

Psychophysics, then, try to prove a theoretical postulate:

1) we may use a correlation between standards of stimulus change and of sensation change in order to quantify sensation. This will allow us to measure sensations along a continuous scale of variations.

But then, we ask, how do we know that there is a correlation? The answer is that

2) we have conducted experiments which show that we have discrete units of sensation that change in a standardizable way with units of a continuously variable stimulus.

So because the stimulus changes continuously, we may standardize the distance between each increment. And because they correlate to discrete units of sensation change, we think that we can treat the sensation units as having a distance between them that corresponds quantitatively with the stimulus differences. However, the experiment did not show that the sensation changes may be treated as though there were any quantifiable distances of sensation between them. We first have to presuppose that there is a continuous interval between sensation levels before we can correlate them in a continuous way with continuously variable stimuli. Hence

3) we base our experiments on the supposition that there is a continuous correlation between sensation alteration and stimulus variation.

Hence altogether:

1) We want to prove that there is a continuous correlation between a) discrete sensation differences and b) continuous stimuli variations so that we may scale and measure sensations.

2) We prove this by means of an experiment.

3) Our experiment supposes that sensation variations are continuous.

Hence the means of proving the theory already presupposes that theory. As Bergson writes,

In a word, all psychophysics is condemned by its origin to revolve in a vicious circle, for the theoretical postulate on which it rests condemns it to experimental verification, and it cannot be experimentally verified unless its postulate is first granted. The fact is that there is no point of contact between the unextended and the extended, between quality and quantity. (70b)

Hence we cannot quantify sensation.

[Next entry in this series.]

[Directory of other entries in this series.]

Images from the pages summarized above, in the English Translation [click on the image for an enlargement]:

Images from the pages summarized above, in the original French [click on the image for an enlargement]:

Bergson, Henri. Time and Free Will: An Essay on the Immediate Data of Consciousness, Transl. F. L. Pogson, (New York: Dover Publications, Inc., 2001).

Available online at:

http://www.archive.org/details/timeandfreewill00pogsgoog

French text from:

Bergson, Henri. Essai sur les données immédiates de la conscience. Originally published Paris: Les Presses universitaires de France, 1888.

Available online at:

http://www.archive.org/details/essaisurlesdonn00berguoft

[Site Topic Directory]