• [cliquer ici pour la version en français]

Dear S. Santini,

this is Frédéric Morain-Nicolier, who is working in image processing and analysis, more specifically on (local) similarity measures between images. In my actual researches, i am thinking about the introduction of results from psychologist (models, laws and experiments). Reading main Tversky's papers and your book (marvelous) "Exploratory Image Databases", it seems there is a problem in one of Tversky's experiments report.

In your book, p. 123, you report the experiment on the asymmetry of similarity between figures:

Each pair had a figure drawn from a set p (for prototypical) and a figure drawn from a set q [...] The p figure had either "better" form than the q figure, or the two had the same goodness of form but the p figure was "simpler" than the q figure.

You conclude some lines later that a majority of subjects selected the statement of the form "q is similar to p".  This leads to the fact that a simpler figure is preferred if the two figures have same "goodness of form". This fact seems in accordance with the prägnanz in gestalt, where a good form is simple, regular, ordered, homogeneous.

My perplexity comes from Tversky's report of the experiment. In "Features of similarity", paragraph "Similarity of figures" :

In the first set, one figure in each pair (denoted p) had better form than the other (denoted q). In the second set, the two figures in each pair were roughly matched in goodness of form, but one figure (denoted p) was richer or more complex than the other (denoted q).

In another paper, "Studies of similarity", paragraph "Study 3: Similarity of figure", Tversky writes:

[...] we expected a "good figure" to be more salient than a "bad figure," although the latter is generally more complex. For pairs of figures that to not vary much with respect to goodness of form, however, the more complex figure is expected to be more salient.

It seems there is a quite disagreement between Tversky's report and your's. My own thinking leads me to be in agreement with your point. But i find strange that Tversky's could be mistaken, with the same mistake in two papers. Do you have ever noticed this disagreement? If yes how do you interpret it?

I look forward to hearing from you soon.

Yours sincerely, Frédéric Morain-Nicolier