These were the first studies I ran during my Ph.D. with Ian McLaren, so 1995-1996, and the first scientific paper I had accepted. I was 24. I think Ian probably saw this project going more in the direction of perceptual learning research, in that the sharpening of generalization gradients under discriminative training is expected from a common-elements view. To his credit, when Experiment 2 made such an explanation look unlikely, he was happy to go with my decision-mechanism explanation, and was really supportive as I developed the winner-take-all simulations. It was from this initial base that I developed my research theme on decision mechanisms in category learning. I seem to remember it taking about three months to get these simulations working, and Mike Aitken quipping that I hadn’t done any psychology that term. The work was all done on Acorn RiscPCs, a somewhat forgotten cul-de-sac of desktop computing.
Re-reading this paper in 2019, it seeems better written than I remember it. On the other hand, by today’s standards, the empirical science falls a little short. An N of 12-20 per condition for a between-subjects comparison, without any argument the effects should be large, is a shortcoming, as is the concluding from a null with NHST inherent in the interpretation of Experiment 2. I also learned a lot about programming, running, and analysing experiments in this project, and my archives are pretty sparse from these early days. So, overall, I’d recommend a large-scale replication before building a programme of research upon these data.
Looking back, it’s interesting to see how much of my future career is foreshadowed in this paper. Work by Greg Ashby, Lee Brooks, Mark Gluck, Rob Nosofsky, Stephen Lea, and David Shanks, are all mentioned. I almost moved to the US to work with Mark. Stephen was important in my move to Exeter in 2000. Lee Brooks’s work comes up a lot in another of my research themes. Greg became something of adversarial figure in my published work from 2009 onwards. David’s work has probably affected my thinking as much as anyone I’ve worked with.