1. Jordan H., Fallah M., & Stoner G.R. Adaptation of gender derived from biological motion

2. Summary What’s the information in point light displays? Consideration of data and stimuli in Jordan paper –Unmentioned issues and improvements More general issues –Interesting issues adaptation might be used to address

3. Biological motion Point light displays contain dynamic motion information about gender and other dimensions It is often forgotten that they also contain form from motion and static structure information Nikolaus Troje This is neglected in this paper

4. Adapatation Adaptation effects are thought to imply the existence of neurons selective for the adapted feature. Here the idea is to adapt to stereotypically male or female point light walkers and see if gender judgments to subsequent test stimuli are biased in the opposite direction to the adapter More to say about adaptation later

5. 6% bias

6. adapting walkers Notice structural and dynamic differences 0.66 female test walker and 0.66 male test walker0.66 female test walker and 0.66 male test walker Height now equal, but some structural differences remain Adaptation could be driven by structural or dynamic cues or both what about dephased adapting walkers?

7. Slight increase in sensitivity for dephased adapters? Not discussed.

8. Residual biases with dephased adapters Authors refer to “low-level” motion adaptation from dephased adapters. Seems more likely that the residual adaptation is due to gender from structure in the displays. Predictions for dynamic only and structure only adapters –No effect for dephased dynamic adapter –Small effect for dephased structural adapter

9. What does dephasing tell us? What kind of system could still perform with (dynamic) dephased stimuli? –One that was detecting a total motion energy difference between stimuli and using it to categorise –So this (naïve) mechanism is ruled out –A system that relied on relative motion between pairs or triads of points at critical locations (e.g. hips) would be confused by dephasing.

10. Improved Experiment Dynamic only adapters Structure only adapters Combined adapters Dephased dynamic Dephased structural Dynamic and structural test sets Measure size of dynamic and structural contributions and whether effects cross over from dynamic adaptation to structure test and vice versa

11. The improved experiment would address a key question –If the adaptation from dynamic crosses over to structural test stimuli and vice-versa, then we have evidence for a “high-level” gender centre –If adaptation to structural information does not produce an effect on dynamic test stimuli then this would suggest that the gender judgments were based on “low-level” cues –The 1 st result would confirm the central point of the paper, the 2 nd result would refute it

12. General comments It would have been more worthy of Nature if there had been no effect of adaptation. –Adaptation effects imply neural substrate, but what theory would predict that we are able to discriminate gender without neurons or networks of neurons that do the job? –Therefore the basic adaptation phenomena reported here is unsurprising and not very interesting by itself.

13. Better use of adaptation Note that gender is not something that can be perceived directly (like e.g., colour), it can only be inferred by correlation with other attributes –Reasonable to ask if there is a unified gender discrimination area in the brain, or whether independent gender sensitive neurons are located in separate modalities. –If adapting to a male voice biases point light test stimuli this would be informative –What is the cue combination model for gender?

14. More useful adaptation questions Duration* Inter-ocular transfer Storage Retinal or visuotopic coordinates *Test stimuli were brief, so the after effect is probably brief. –But if it can persist this would have implications for social interactions in isolated same sex groups

15. Adaptation and fMRI The authors allude to a potential brain mapping experiment Present a series of male walkers –Ideally they might vary in other dimensions such as size, emotional state, walk vs. run –Then present one female walker Then search brain for voxels whose activation is high at start of series, then declines rapidly, and finally recovers when the female is presented –The deliberate variation in other dimensions helps rule out voxels that are interested in other features.