IV. Motion recognition

Although we showed in the foregoing section that birds can discriminate between different types of movement, it is another thing entirely to say that they discriminate them on the basis of movement cues. A horizontal movement of an object takes it to different parts of visual space than a vertical movement, and it might be this locational difference that makes discrimination between them possible. How could we show that it was movement, itself, not the consequences of movement, that a bird was discriminating?

We have already cited some evidence that argues in this direction. For example, Emmerton's (1986) failure to find directional invariance in most Lissajous figure discriminations argues strongly that it is the motion of the dot, not its trajectory, that is being discriminated. Similar Cook and Katz's (1999) evidence that moving objects were discriminated better than the same objects at rest implies that motion contributes information.

However, the strongest demonstration that discrimination between types of movement, and especially complex types of movement, can be achieved using movement cues alone is to replace with stimuli with representations based on a handful of points of light, the arrangement Johansson (1973) described as a "biological motion" display. Dittrich et al. (1998, Experiment 3) showed that pigeons could be trained in movement category discriminations using scenes of pecking and walking movements by other pigeons presented under point-light conditions. Ryan, Lea, Alklind, and Dittrich (2001) extended this result to chickens, using the identical stimuli as in the pigeon study, of which two examples of Dittrich et al's full-detail (pecking / walking) and point-light (pecking / walking) displays can be viewed here.

Eleven of 16 birds reached criterion, with a final performance of 80% correct within 27 sessions and the remaining birds reached at least 60% within 30 sessions. Fifteen of the birds were then transferred to a second discrimination in which the alternative kind of stimuli were used. In half of the cases compatible transfer (full-light walking stimuli replaced by point-light walking stimuli) was tested and in the remaining cases incompatible transfer was used (full-light walking stimuli replaced by point-light pecking stimuli). Compatible transfer results were better  than incompatible one indicating that indeed discrimination transfer between full-light and point-light information seemed possible. Movement cues alone seemed sufficient to enable a discrimination between moving birds.

Furthermore, we can show that such pure motion cues must play a part in discrimination even of full detail displays. In Dittrich et al.'s Experiment 2, birds that had been trained to discriminate pecking and walking with full-light video scenes were then tested with point-light scenes of the same movement categories. There was some transfer of the discrimination. Using a slightly different testing technique, Lea et al. also obtained transfer between point-light and full-light displays. They tested both for transfer from full-detail to point-light, and for the reverse, and found only slight differences in the extent of transfer between the two. Transfer to full-light displays seem easier. 

Next section: Motion concepts