21 Jan Memory-based formation of preferences
Why do we prefer what we prefer? Take cereal bars. Do I genuinely prefer an almond bar over one with hazelnuts most of the time? Or does this almond bar look like the one I had the other day, which was delicious, and happend to have almonds? (I personally prefer chocolate, but that’s a different story.)
[column3]
View one: People prefere things, because they like the attributes of some things better than the attributes of other things — let me call this the classic view on preferences, because it has been around for long (L. Keeney & Raïffa, 1976). What this means is that I prefer the almond bar because I really prefer almonds to other nuts. People, following this view, trade off attributes they like against attributes they dislike. Confronted with an almond-raisin bar, maybe I love almonds and dislike raisins. People should, in the end, prefer the thing where the sum of attributes is most attractive to them. [/column3]
[column3]
The newer view: People prefer something because of similarity to experience. Options remind them of past things they really liked. This means I prefer the almond bar because it resembles yesterday’s almond cereal bar, which I liked. This is interesting because it links preferences to experience and memory, such wonderfully fundamental cognitive processes. On a theoretical note, the view relates to case-based choice (Itzhak Gilboa & David Schmeidler, 2001), which again relates to classic psychological processes in classification (Nosofsky, 1986). (I’ve also published on classification, it’s fascinating). [/column3]
[column3 last=yes]
Does the memory-based view hold true? We rigorously test the memory-based view against the attribute-valuation view using two incentivized preferential choice tasks. In one task, you select pens. In the other task you taste cereal bars. We then use computational cognitive modeling at the aggregate and individual level to get at the answer. The results clearly favor one theory, and will hopefully soon be published. I don’t spoil the results because I have no idea if this keeps me from publishing the paper (and my co-author was not in favor of a pre-print. I do respect his views). Let me update this post, once the manuscript is accepted.
[/column3]