How to Adjust the Response Window

The temporal characteristics of the response window may be adjusted from block to block depending on a given subject's performance. The way the script specifies the response window method also determines the adjustment procedure.

The default window adjustment procedure is:

The response window is moved back 33 ms for subsequent blocks if

1. the percent of correct responses for the block <= 55%.
   
   
2. the percent of correct responses for the block <= 65% and mean latency is over 100 ms greater than the current window center.

The response window is moved forward 33 ms for subsequent blocks if

1. the percent of correct responses for the block >= 80% and mean latency is no more than 100 ms greater than the current window center.

Using the <response> element, it is possible to customize various aspects of the window adjustment procedure, including conditions for incrementing the window center (moving it back), conditions for decrementing the window center (moving it forward), the increment and decrement amounts, the minimum and maximum window center values, and whether to base the adjustment algorithm on mean or median latency.

Response Window for Blocks

The response window procedure can be specified at the block level, as in the following:

<block myblock>
/ responsemode = window(100, 100, windowstim)
</block>

In this example, adjustment of the window center for myblock is independent of the subject's performance on different blocks defined in the script. Each adjustment to the response window of myblock affects only subsequent runs of myblock and is unaffected by performance on blocks with other names.

Response Window for Expt

The response window procedure can be specified at the expt level, as in the following:

<expt>
/ responsemode = window(100, 100, windowstim)
</expt>

In this case, the scope of the window center is the experiment. All blocks within the experiment will share the experiment's window center, with the exception of blocks that have the response window explicitly defined for themselves (see above). Each block that uses the shared experiment window center may adjust that center based on the subject's performance, and subsequent blocks in the experiment will use the adjusted window center.

Response Window for the Response Element

The response window procedure can also be specified within a response element, as in the following:

<response myresponse>
/ mode = window
/ rwhitduration = 200 
/ rwhitstimulus = hitstim
/ rwcenter = 400
/ rwwidth = 300
/ rwstimulus = windstim
</response>

In this case, the myresponse element has its own response window center, and all block or expt elements that specify "/response = myresponse" will share that window center. This window center may be adjusted based on performance of any block that uses myresponse, and the adjusted window center will be used by subsequent blocks that use myresponse.