An Analysis of the Blind Variation and Selective Retention Theory of Creativity

Abstract

Picasso's Guernica sketches continue to provide a fruitful testing ground for examining and assessing the Blind Variation Selective Retention (BVSR) theory of creativity. Nonmonotonicity—e.g., as indicated by a lack of similarity of successive sketches—is not evidence of a selectionist process. Although the notion of blindness originally implied randomness, it now encompasses phenomena that bias idea generation, e.g., the influence of remote associations on sketch ideas. However, for a selectionist framework is to be applicable, such biases must be negligible, otherwise evolutionary change is attributed to those biases, not to selection. The notion of variants should not be applied to creativity; without a mechanism of inheritance, there is no basis upon which to delineate, for example, which sketch ideas are or are not variants of a given sketch idea. The notion of selective retention is also problematic. Selection provides an explanation when acquired change is not transmitted; it cannot apply to Picasso's painting (or other creative acts) because his ideas acquired modifications, as he thought them through, that were incorporated into paintings and viewed by others. The generation of one sketch affects the criteria by which the next is judged, so sequentially generated sketches cannot be treated as members of a generation, and selected amongst. Although BVSR is inappropriate as a theoretical framework for creativity, exploring to what extent selectionism explains the generation of not just biological form but masterpieces such as Picasso's Guernica is useful for gaining insight into creativity.

Acknowledgments

I would like to thank Stefan Leijnen for comments on the article. This work was funded in part by grants from the Social Sciences and Humanities Research Council of Canada and the Concerted Research program of the Flemish Government of Belgium and the Free University of Brussels.

Notes

¹The term quasi-random is avoided here because—like quasi-infinite or quasi-bottomless—it is not clear what it means (although I can think of a meaning for quasi-topless). The word pseudo-random, on the other hand, has been clearly defined. It refers to a sequence that is not truly random because it was generated by a deterministic process (for example, by a computer), but that exhibits statistical randomness and, therefore, cannot be distinguished by humans from random. (Because it is not genuinely random, it can be used repeatedly to produce the exact same sequence of numbers.) If the generation of ideas were pseudo-random, that would be sufficient, because the critical factor is whether it can be approximated by a random distribution. To the extent that the distribution of traits or characteristics deviates from a random distribution, natural selection gives a distorted model. Change over time is attributable to the source of this deviation, not to natural selection. However, the generation of ideas is by no means pseudo-random; it is biased by associative memory, drives, the present situation, and so forth. Therefore, it cannot be approximated by a random distribution, and change over time cannot be modeled by natural selection.

²Actually, in some biological situations, such as assortative mating, the assumption of randomness does not hold, and in such cases natural selection is not an appropriate model.

³Nevertheless, there are logistic reasons why change at the individual level cannot be highly nonmonotonic. One has to do with epistasis, wherein what allele is best at one locus depends on what allele is present at another locus. If the mutation rate is too high, evolution grinds to a halt because change at one locus affects not just that locus but the stable arrangement it has found with other loci to which it is epistatically linked. The bottom line is that nonmonotonicity in the creative process does not indicate that creativity is Darwinian.

⁴Note that mapping the idea to the genes is no less problematic than mapping the idea to the individual organism. This can be illustrated with a simple example. Say we genetically modify the iris of your eye such that it produces different pigment and your eye looks brown rather than blue. The probability that your offspring have the genes for blue eyes will still be as high as it was before this procedure. Blue lives on in this lineage. But consider the situation in which, in the course of painting a painting, you paint over a blue blob with brown paint. There is no mechanism in place that causes the painting to revert back to having the blue blob in it; i.e., that prohibits the brown paint from being present in future iterations of the painting, or from future paintings done in that style. Unless you actively scrape it off, the brown paint will continue to hide the blue blob. Blue is lost from the lineage. The reason for this is that the underlying mechanism of change in the two situations is very different (for further analysis, see Gabora, 2004, 2008).