Behavior Visualizer

[RoboDoig]

• Behavior Visualizer (1)
  • Intended to show events: ManualWater/Choice/Rewarded/NoResponse with R/L y-axis
• Behavior Visualizer (2)
  • Intended to show history of: p_L, p_R, p_R_frac, choice_frac, reward_frac, finish_frac

In general with these event visualizers, want the option of toggling hide/show of certain streams
If possible, these visualizers should be generic, configure plot type and available streams

[RoboDoig]

@micahwoodard what do the black bars on the top plot represent? I don't see their correspondence in the legend.

[micahwoodard]

@micahwoodard what do the black bars on the top plot represent? I don't see their correspondence in the legend.

I am actually not 100% sure. @ellahiltonvano are these black bars licks?

[ellahiltonvano]

@micahwoodard yes, those are licks

[RoboDoig]

@ellahiltonvano are they intended to show just whether there was a response lick during a trial? or actual timing of licks within the trial? I don't think there is any lick data in the trial outcome data class other than spout choice / reward.

Another question, in the mockup the 'no-choice' markers are split between the L/R parts of the y-axis, although from my understanding they can't really be classified as left vs. right as there is no lick present in these cases.

[ellahiltonvano]

@RoboDoig they show actual timing of all licks. @alexpiet can you confirm?

[alexpiet]

@RoboDoig

• Yes, the black tick marks are the timing of all licks, showing the actual timing within the trial.
• the no-choice (we call them "ignore") ticks are meant to be exactly in the middle between L and R, since they aren't classified as left or right.

[RoboDoig]

@alexpiet - OK, that might be slightly tricky as lick information is not contained in the TrialOutcome data class, unless something has changed. Given that the x-axis of this plot is trial index would this mean rescaling lick timing to a normalized timebase so that it can be plotted on a trial index axis?

[alexpiet]

The x-axis is time, not trial index (confusing that the label is cut-off in the image).

[RoboDoig]

Ah ok I see, apologies I wasn't clear on this as the example implementation in the repo appeared to be trial index based.

[micahwoodard]

@RoboDoig some context in case it is useful, the desired behavior visualizer is a graph implemented in this repo https://github.com/AllenNeuralDynamics/dynamic-foraging-task/blob/develop/src/foraging_gui/Visualization.py. It is what is currently being used by the team to assess behavior. Instead of the current visualizer Bruno implemented, the team wants a recreation of what is here.

[alexpiet]

If using time based indexing, and plotting every lick is a big burden. We could use trial based indexing and only plot if there is a response or not.

[RoboDoig]

That's how I've implemented it for now, I will also look into the time based indexing though.

[RoboDoig]

@micahwoodard are you happy for me to close the issues in the visualizer refactor milestone following PR #139?