Empirical Evidence Against My View of Dream Reports

Nowadays, most Americans report dreaming in color. In the 1950s, most Americans reported dreaming in black and white. In a series of articles I have argued that the reason for this change is not that people used to dream in black and white and now dream in color. Rather, I argue, people over-analogize dreams to movies. Thus, as movie technology shifts, people's dream reports shift, though their dreams themselves remain the same.

(Two pieces of evidence for this view: (a.) The use of color terms ("brown", "orange", etc.) in dream diaries seems to have been consistent since the 1950s. (b.) Color dream reporting correlates with group history of black-and-white media exposure across socioeconomic groups in China.)

A new study by Hitoshi Okada and colleagues in Japan calls my research into doubt. In 1993, Okada and colleagues had found that young Japanese respondents tended to report colored dreaming while older respondents tended to report not dreaming in color -- a result entirely in accord with my hypothesis, due to respondents' presumably different histories of black-and-white vs. colored media exposure. Now in 2011, Okada at al. find almost exactly the same pattern of responding. Thus, the cohort of respondents that was in their 20s and 30s in 1993, and who reported mostly colored dreaming back then, reports relatively infrequent color dreaming now. Twenty years of (presumably) colored media exposure appears not to have shifted them toward reporting more colored dreaming -- if anything, the opposite.

Maybe these results can be reconciled with my view. For example, maybe older Japanese regard as the archetypal movie the old-fashioned high-art black-and-white movies of Kurosawa and others. But that doesn't seem especially likely.

Another possibility (as always!) is that Okada's research is open to interpretations other than its face-value interpretation.

The following is Okada et al.'s entire description of their questionnaire:

The participants were required to check one of five categories describing the frequency with which color occurred in their dreams during the past year: 1 (always), 2 (sometimes), 3 (occasionally), 4 (seldom), or 5 (never) (p. 216).

In English, I don't know that "sometimes" implies higher frequency than "occasionally", but I trust that this is just an infelicity of translation from the original Japanese.

One worry is that this measure has no denominator. So here's one possible explanation of the Okada et al. results: Older Japanese people report dreaming less in general than do younger Japanese, so they report less frequent colored dreaming too. This would be consistent with their self-reported ratio of black-and-white to colored dreaming being about the same. (In my own work on the issue, I ask some respondents about absolute frequency or colored dreaming and others about the proportion of colored to black-and-white dreams.)

Another potential concern is non-response bias. Okada et al. state that their participants were "students in Bunkyo University, Jissen Women’s University, and Iwate University, or members of their families" (p. 215). They don't indicate the response rates of the family members, but it's possible that only a minority of family members who heard about the questionnaire chose to respond. If so, those family responders would mostly be people with higher-than-average interest in the issue of black-and-white vs. colored dreaming. And we might reasonably worry that such people would not have views on that question that are representative of the population as a whole. (This is, of course, the notorious problem with online polls.)

I'd be very interested to see a follow-up study addressing these concerns.