A Combined Perceptual, Physico-Chemical, and Imaging Approach to ‘Odour-Distances’ Suggests a Categorizing Function of the Drosophila Antennal Lobe
Figure 3
Concordance of perceived distance across four types of recognition experiment.
(A) Re-presenting the data from Figure 2B, pooled for odour pairs. The stippled grey line represents the learning indices that were found when TRAINing and TESTing odour were identical (from Fig. 2A). To the extent that flies regarded the TESTing odour as different from the TRAINing odour, learning indices should approach zero; thus, the degree to which flies regarded both odours as different can be quantified by the Perceptual Distance Score 1 (red arrows). In (A′) these scores were presented normalized to the highest median score thus obtained. Sample sizes are from left to right: 32, 32, 32, 32, 31, 32. (B) Same as in (A), except that a 180-min break was given between training and test. Sample sizes are from left to right: 24, 24, 24, 24, 24, 24. (C) Flies were trained with a given odour, and then were tested for their choice between that trained odour versus a novel, not previously trained odour. Thus, if the flies regarded the two TESTing odours as the same, scores should be zero. To the extent that both odours, however, were regarded as different by the flies, learning indices should increase. The level of perceived difference thus can be approximated by the Preceptual Distance Score 3 (green arrows). In (C′) these scores are presented normalized to the highest median value thus obtained. Sample sizes are from left to right: 24, 24, 20, 23, 24, 24. (D) Flies were trained such that one odour was punished but the other odour was not punished; then, flies were tested for their choice between these two odours. Thus, if the flies could not tell the two odours apart, scores should be zero. To the extent that both odours, however, could be discriminated by the flies, learning indices should increase. The level of perceived difference thus could be approximated by the Preceptual Distance Score 4 (blue arrows). In (D′) these scores were presented normalized to the highest median value thus obtained. Sample sizes are from left to right: 15, 11, 12, 11, 11, 12. * and ns refer to P<>0.05 in Kruskal-Wallis tests. Other details as in Fig. 1.