Findings

 

This page provides a brief summary of the major findings of the project. More detailed reports of the experiments can be found in the published outputs of the project, as listed on a separate page.

Adaptation to implied motion

The first experiment we conducted was a replication of Winawer et al.’s (2008) implied motion after-effect (IMAE), using our stimuli and methods (adaptation to locomotion still-frames; left vs. right responses to drifting test patterns).

To test whether our IMAE was due to sensory bias or to response/decision bias, we compared results using two tasks, one involving right vs. left responses and the other involving top vs. bottom responses (see the Methods page). We obtained a IMAE using the left vs. right task, but not using the top vs. bottom task.

On the basis of these results we concluded that the IMAE is not due to low-level adaptation but to response/decision bias.

Adaptation to human locomotion

We conducted a series of experiments on the apparent speed of human locomotion, and found:

Basic effect (1): Adaptation to slow-motion (0.48x playback) walking causes standard-speed walking (1x playback) to appear too fast, shifting the P50 point towards slow-motion speeds. Similarly adaptation to fast-forward (1.44x playback) walking causes standard-speed walking to appear too slow, shifting the P50 point towards fast-forward speeds. The same effects were obtained with running videos.

Cross-adaptation (2): Adapting to walking  videos and testing with running videos, and vice-versa, assessed whether the effect in (1) was due to adapting speed expressed relative to standard-speed, or to the retinal speed of the adapter. Results showed that the adaptation is driven by retinal speed.

Row-scrambled adaptation (3): Adaptation to row-scrambled running,  and testing with intact running, assessed whether the effect in (1) depended on the presence of recognisable human forms during adaptation. We found a significant effect of adaptation, though the effect was somewhat smaller than that  in (1), indicating that recognisable forms are not necessary to obtain the effect, but they do have some influence.

Retinotopic specificity (4): Results in (3) imply a low-level site of adaptation, as obtained in previous experiments on velocity adaptation (velocity after-effect or VAE). The VAE does not transfer across different retinal locations. We tested whether the adaptation in (1) also transfers across locations. Results showed that the effect did transfer across locations, consistent with a relatively high-level site of adaptation.

Velocity after-effect or VAE (5): As a second way to test whether the effect is consistent with a low-level VAE, we measured the standard low-level VAE obtained from adapting and testing with our stimuli (row-scrambled images as used in (3)). Results were consistent with previous reports of the VAE, not with the results of this project in (1). We concluded that our adaptation effect cannot be explained by low-level velocity adaptation.

Column-scrambled adaptation (6): Previous research (e.g. Burr and Ross, 1982) has suggested that image temporal frequency content (flicker rate) plays a role in visual estimation of objective velocity. To test whether temporal frequency content could contribute to adaptation to locomotion speed, we conducted an experiment in which the adapting stimulus was a column-scrambled version of the running video used in previous experiments. Column scrambling removes information about stimulus direction but preserves the local flicker properties of pixels. We still obtained an adaptation effect similar to that found in (3) but smaller than that found in (1). We concluded that a component of adaptation to locomotion speed may be mediated by changes in the relative gain of visual channels tuned to temporal frequency. This kind of process has been proposed as the substrate for re-normalisation effects in perception which serve to maintain perceptual constancies such as colour constancy. However the smaller effects obtained using scrambling manipulations also point towards a contribution from high-level processes as well.

The effect of TMS on PLW perception (7): A collaborative study with colleagues in Italy tested whether perception of locomotion in point-light walkers (PLWs; see Methods) is mediated by the low-level cortical area (MT) which is involved in the perception of simple translational motion. TMS applied to MT interfered with low-level motion discrimination but did not interfere with PLW perception, indicating that PLW perception involves higher level processing areas.

Adaptation to point-light walkers (8): As a final test for high-level mediation of adaptation to locomotion speed, and given the result in (8), we tested for an effect using PLWs as adapting and test stimuli. Based on the results of other experiments in the project showing evidence for a high-level locus of adaptation, we predicted that PLWs should be effective adapting stimuli. Using the same MSS technique as employed in other experiments, we obtained an adaptation effect using PLWs that was at least as strong as those obtained using videos of real runners. To test whether response bias contributed to the measured adaptation, we also conducted an experiment using a 2AFC task as described in Methods: After adaptation participants reported which of two PLW test stimuli appeared to move at a speed closer to natural locomotion speed. We found a significant effect of adaptation using this task, though it was smaller than the effect obtained using MSS. We concluded that a component of adaptation is sited at the high-level processes involved in perception of PLWs, and that decision or response bias may also contribute to the obtained effect.