Extend catch trial definition

[dougollerenshaw]

• adds functions for calculating response and d' matrices

• adds a function called annotate_stimuli to utilities.py that annotates all stimuli in stimulus_presentations with the following columns: 'trials_id': the corresponding ID of the stimulus in the trials table 'previous_image_name': the name of the stimulus on the last flash 'next_start_time': The time of the next stimulus start 'auto_rewarded': True for trials where rewards were delivered regardless of animal response 'trial_stimulus_index': index of the given stimulus on the current trial 'response_lick': Boolean, True if a lick followed the stimulus 'response_lick_times': list of all lick times following this stimulus 'response_lick_latency': time difference between first lick and stimulus 'previous_response_on_trial': Boolean, True if there has been a lick to a previous stimulus on this trial 'could_change': Boolean, True if the stimulus met the conditions that would have allowed to be chosen as the change stimulus by camstim:

  • at least the fourth stimulus flash in the trial
  • not preceded by any licks on that trial

• modifies the get_behavior_stats function in utilities.py to use a new "stimulus_based" definition of catch stimuli, based on the above annotations. if 'stimulus_based' (default), calculates hit and false alarm rates using every stimulus on which a change could have occurred. This dramatically increases the number of stimuli used when calculating the false alarm rate, reducing the noise on the false alarm estimate.

[alexpiet]

• adds functions for calculating response and d' matrices
• adds a function called annotate_stimuli to utilities.py that annotates all stimuli in stimulus_presentations with the following columns: 'trials_id': the corresponding ID of the stimulus in the trials table

you mean the ID of the trial?

'previous_image_name': the name of the stimulus on the last flash

What does this do during omissions?

'next_start_time': The time of the next stimulus start

Again, what does this do during omissions?

'auto_rewarded': True for trials where rewards were delivered regardless of animal response 'trial_stimulus_index': index of the given stimulus on the current trial

how is this different from trials_id?

[dougollerenshaw]

@alexpiet Yes, the ID of the trial on which the stimulus occurred. I changed the docstring to read: 'trials_id': the corresponding ID of the trial in the trials table in which the stimulus occurred

It will list 'omitted' if the last stimulus is omitted. I changed the docstring to read: 'previous_image_name': the name of the stimulus on the last flash (will list 'omitted' if last stimulus is omitted)

This also applies to omissions. I changed the docstring to read: 'next_start_time': The time of the next stimulus start (including the time of the omitted stimulus if the next stimulus is omitted)

This is the index of the stimulus within the trial. For example, the first stimulus in a trial has index 0, the second stimulus in a trial has index 1, etc. This is useful for determining where the stimulus would have been in the geometric distribution. Stimuli with indices 0-3 could not have been change stimuli. I changed the docstring to read: 'trial_stimulus_index': index of the given stimulus on the current trial. For example, the first stimulus in a trial has index 0, the second stimulus in a trial has index 1, etc