Reader Comments

Post a new comment on this article

motion sensitivity improvement might instead be observer bias

Posted by perceptionsydney on 08 Mar 2007 at 05:25 GMT

Figure 3 purports to show that motion sensitivity improved for the paired direction but not for the non-paired directions. However, we believe that this pattern of change in percent correct might instead by explained by an increase in observer bias to report the paired direction. Certainly the increase in percent correct for the paired direction could be explained this way. To address this, we suggest that the authors calculate d' (sensitivity) for the paired direction. They could get a measure of false alarms by taking the number of paired-direction responses to all the unpaired directions.

RE: motion sensitivity improvement might instead be observer bias

aseitz replied to perceptionsydney on 09 Mar 2007 at 03:06 GMT

We calculated d' for this data and confirmed that sensitivity improved significantly for the paired direction (P < 0.005, ANOVA).

Note: In other studies looking at task-irrelevant learning for motion directions paired with RSVP targets we have also found sensitivity improvements (e.g. Seitz and Watanabe Nature, 2003; Seitz, Nanez, et al, PNAS, 2005; Seitz, Lefebvre, et al, Current Biology, 2005).