Adding second stimulus #1276

XX-Yin · 2025-01-05T00:27:33Z

Pull Request instructions:

Please follow the update protocol
Answer the questions below in detail. Your responses will be emailed to experimenters.
If the experimenters must do anything new, provide detailed step by step instructions on the wiki
If computer maintainers need to manually update anything, provide detailed step by step instructions
Use markdown syntax in order for your comments to be rendered reliably in the email: "1." instead of "1)", use four spaces for indents.
If you use the keyword "skip email" in the title, it will skip the email updates
Merges from "develop" into "production_testing" should use the keyword "production merge" in the title for reliable indexing of updates
Merges from "production_testing" into "main" should use the keyword "update main"

Describe changes:

Add a second auditory stimulus.
Differentiate the auditory cues for rewarded and non-rewarded trials. The rewarded auditory cue is stored at index 5 of the sound card, while the non-rewarded auditory cue is stored at index 6. If there is no response, no sound will be played.
Introduce a delay to control the timing between the reward outcome (reward or no reward) and the second cue.
Replace the fixed reward delay with a distribution (exponential or uniform). If the second stimulus is enabled, the reward delay corresponds to the second "go" cue. Otherwise, it corresponds to the reward outcome.

What issues or discussions does this update address?

#1263 (comment)

Describe the expected change in behavior from the perspective of the experimenter

This update aligns with our current pipeline, but a few adjustments are necessary:

Adding the following variables to the NWB file:
- B_SecondStimulusDelay: The second stimulus delay sampled for each trial.
- B_GoCueTimeSoundCardSecondStimulus: The harp timestamp of the second sound.
- B_RewardDelay: The reward delay time sampled for each trial.
- TP_GiveSecondStimulus: A parameter indicating whether the GiveSecondStimulus is turned on or off.
- TP_RandomnessSecondStimulus: A parameter defining the random sampling method used for the second stimulus delay.
- TP_SecondStimulusBeta: A parameter specifying the beta value used for the exponential distribution of the second stimulus delay (not applicable for uniform distribution).
- TP_SecondStimulusMin: A parameter specifying the minimum value for the second stimulus delay.
- TP_SecondStimulusMax: A parameter specifying the maximum value for the second stimulus delay.
- TP_RandomnessRewardDelay: A parameter defining the random sampling method used for the reward delay.
- TP_RewardDelayBeta: A parameter specifying the beta value used for the exponential distribution of the reward delay (not applicable for uniform distribution).
- TP_RewardDelayMin: A parameter specifying the minimum value for the reward delay.
- TP_RewardDelayMax: A parameter specifying the maximum value for the reward delay.
Deleting the following variables to the NWB file:
- TP_RewardDelay
Adding the following variables to the curriculum
- GiveSecondStimulus: Default is off if not set from the curriculum
- RandomnessSecondStimulus: Default is Even if not set from the curriculum
- SecondStimulusBeta: Default is 2
- SecondStimulusMin: Default is 1
- SecondStimulusMax: Default is 8
- RandomnessRewardDelay: Default is Even
- RewardDelayBeta: Default is 2
- RewardDelayMin: Default is 1
- RewardDelayMax: Default is 8
Deleting the following variables from the curriculum
- RewardDelay
Uploading waveform to index 5 and 6 of the soundcard to use the second stimulus.

Describe any manual update steps for task computers

Was this update tested in 446/447?

This PR is a prototype for adding a second stimulus, and it will be handled over to @rachelstephlee @hagikent for full testing, updates and modifications. @alexpiet @hanhou @bruno-f-cruz @micahwoodard for sanity check.

hagikent · 2025-01-06T23:09:27Z

Great! Thanks @XX-Yin
I will check the changes in the bonsai side later today. I cannot critically read GUI side, so better letting @micahwoodard

Uploading waveform to index 5 and 6 of the soundcard to use the second stimulus.

I will take care of this too.

bruno-f-cruz · 2025-01-07T03:37:46Z

Here's the refactored version (#1286).
I focused on the snippet you sent above without changing anything else. So I hope I understood the scope of your changes.
It fixes a likely bug you might have encountered when testing where the SelectMany would not close on each OnNext event (try to close the inner sequence, if you want to be explicit about it, just use a simple Take(1) in most cases). If I am understanding the logic correctly, I think the board would still play the sound but you would progressively send more and more messages per trial. This would likely increase the playback of the sound by 1-2ms per trial which could become a problem at some point. It also refactors a bit of the logic to make it easier to maintain in the future should you need to. If you have problems let me know.

I have also refactored the workflow to make it easier to read and maintain in the future should you require. In general I advise abstracting the functionality as much as possible from the hardware (e.g. Coding the tones via subjects, and converting the logic into hardware instructions., i.e. HarpMessages, as late as possible). This makes code more readable for anyone trying to maintain it but also easier to refactor in the future should you need to use something else to deliver a tone.

If you find bugs or have questions, lmk!

XX-Yin · 2025-01-07T06:33:18Z

@bruno-f-cruz Thank you! This looks great! @hagikent and @rachelstephlee will test your updates.

I think you're likely right that "SelectMany would not close", but during testing, I didn't observe more and more messages. Could you show me sometime? I'm really curious about this.

rachelstephlee · 2025-01-07T18:31:01Z

thank you everyone for getting this done so fast! our meeting on thursday will end early if we dont' need to do much (i hate long meetings).

see you in a few days!

rachelstephlee · 2025-01-08T05:06:21Z

In preparation for Thursday, Kenta and I will try to figure out the starting parameters we will want to use for this set up. Please let us know if there's anything else we can prepare ahead of time for our Thursday meeting so we can have a quick and efficient meeting.

So far, the parameters we will need to provide are:

tone sounds for CS+/CS-
timing between key events (go cue -> lick choice -> CS+/CS- -> new trial etc).

XX-Yin · 2025-01-08T19:03:02Z

a little more detail about parameters in the GUI if you plan to test it. Related parameters are shown below. And the second stimulus should be selected as on.

hagikent · 2025-01-08T22:56:29Z

@XX-Yin
Thanks! I suppose beta/min/max are for exponential dist? Is the distribution generated on the GUI side or Bonsai-side?
@bruno-f-cruz the other day showed me how to flexibly generate different distribution in Bonsai including constant values. Can we adopt it?

We can discuss this tomorrow, thanks!

XX-Yin · 2025-01-08T23:04:40Z

@hagikent
The distribution is controlled by the randomness parameter. If it's selected as the Exponential, beta is involved. If it's selected as even, then only use min and max will be used. Currently, the randomness parameter also controls the random distribution of the ITI and delay. You can add an independent randomness control for the second stimulus if you want.

The distribution is generated on the GUI side.

hagikent · 2025-01-09T03:36:32Z

thanks for the clarification @XX-Yin !
in a long-run, we might want to let bonsai to take care of rand-get. let's discuss tomorrow.

hagikent

A comment on dist selection.

Also make the default parameters for the second delay/reward delay more realistic!
(credit assignment won't work with ~8s reward delay!)

hagikent · 2025-01-09T04:27:11Z

src/foraging_gui/MyFunctions.py

@@ -392,16 +395,23 @@ def _generate_next_trial_other_paras(self):
        # get the ITI time and delay time
        if self.TP_Randomness=='Exponential':


Let's make distributions independently controlable.
-ITI needs to be exp so that it gives near-flat hazard.
-@rachelstephlee prefers normal dist for CS/Reward delays

I no longer prefer normal distributon-- @alexpiet made a good point it doesn't have to be normal, exponential OK also. it shouldn't matter too much.... hopefully this is a small change we can change as we test how the mice responds. The rest would be guess-work and depend on how well model-fitting works after the fact.

XX-Yin · 2025-01-09T20:51:49Z

I didn't see repeated messages, there were 5 trials with 5 messages to the soundcard. However, I added take 1 to the SelectMany for extra safety.

hagikent · 2025-01-09T21:26:58Z

ah, @XX-Yin are you directly modifying this branch?
I thought, first merging @bruno-f-cruz 's bc branch into yours, and then adding minor fix would be the strategy.

I'm ok if the former is easier for you! Just for clarification.

XX-Yin · 2025-01-09T21:44:50Z

ah, @XX-Yin are you directly modifying this branch? I thought, first merging @bruno-f-cruz 's bc branch into yours, and then adding minor fix would be the strategy.

I'm ok if the former is easier for you! Just for clarification.

I think we can go with the current branch without merging Bruno's branch.

hagikent · 2025-01-09T21:48:42Z

Ok, sounds good.
Hope @micahwoodard can review the modification in the GUI side, in particular, newly added parameter fields are correct.

@XX-Yin
Also, you mentioned we need to update nwb packaging, correct? Can you unpack it a bit? The pipeline needs upgrade?

XX-Yin · 2025-01-09T21:53:17Z

@hagikent
I will give a detailed description about changes we need to make on the NWB packaging and the Curriculum after adding two extra parameters to independently control the distribution of the second delay and the reward delay.

XX-Yin · 2025-01-13T20:11:34Z

@hagikent @rachelstephlee I added the independent randomness control to the second stimulus and the reward delay. I also described the updates required for the downstream pipeline in the pull request comment.

rachelstephlee · 2025-01-27T00:39:43Z

Thank you so much Xinxin!
I assume Kenta will test this new version out in 428 and then we can merge. @hagikent, please invite me when you test it in person, i'd like to see that happen in person this time. tysm.

hagikent · 2025-01-27T02:41:57Z

I already did dry run at 428 3-weeks ago, and the new CS+/- function works. So, from my side, it's ok to merge.
As commented above, however, I'm not qualified to review the change in the GUI side. Need help/review from @micahwoodard @alexpiet @hanhou . We can also merge it and test it during the integration testing if it won't affect "normal" foraging functions.

@rachelstephlee
Let's try with a few living mice before long!

hanhou · 2025-01-27T16:38:27Z

We should update the nwb transfer script to accommodate the new version (while keeping it compatible with old versions) before merging this to main. Otherwise it may immediately break all downstream pipelines and AutoTrain. We should finish these steps:

Nwb conversion
(Given that the effort to centralize nwb conversion is still in progress, I'm listing all three places here)
@hagikent could you share a sample behavior json from your test run here? @XX-Yin could you use it to update the nwb transfer script? I can help you test it.
@Ahad-Allen please make sure to update the nwb transfer script on your end.
@ZhixiaoSu please make sure to update the nwb conversion in your ephys pipeline, if necessary
AutoTrain
@hanhou will make sure it doesn't break our current AutoTrain system
@micahwoodard please update these new parameters in your new task schema

rachelstephlee · 2025-01-27T17:08:32Z

Yes absolutely. Thank you for reminding us of these important TODO’s. We definitely don’t want to break things downstream for other folks. Given that @ahad is in the middle of consolidating the behavior -> nwb script, it means we may want to wait until he finishes that before we merge altogether. That way, we won’t have to do a separate fix for Ephys, and just need to update one script. From: Hou, Han ***@***.***> Date: Monday, January 27, 2025 at 8:38 AM To: AllenNeuralDynamics/dynamic-foraging-task ***@***.***> Cc: Rachel Lee ***@***.***>, Mention ***@***.***> Subject: Re: [AllenNeuralDynamics/dynamic-foraging-task] Adding second stimulus (PR #1276) We should update the nwb transfer script to accommodate the new version (while keeping it compatible with old versions) before merging this to main. Otherwise it may immediately break all downstream pipelines and AutoTrain. We should finish these steps: * Nwb conversion (Given that the effort to centralize nwb conversion is still in progress, I'm listing all three places here) * @hagikent<https://github.com/hagikent> could you share a sample behavior json from your test run here? @XX-Yin<https://github.com/XX-Yin> could you use it to update the nwb transfer script? I can help you test it. * @Ahad-Allen<https://github.com/Ahad-Allen> please make sure to update the nwb transfer script on your end. * @ZhixiaoSu<https://github.com/ZhixiaoSu> please make sure to update the nwb conversion in your ephys pipeline, if necessary * AutoTrain * @hanhou<https://github.com/hanhou> will make sure it doesn't break our current AutoTrain system * @micahwoodard<https://github.com/micahwoodard> please update these new parameters in your new task schema — Reply to this email directly, view it on GitHub<#1276 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ADSAQJ6EBIVWTMAZQNMJXAD2MZOJVAVCNFSM6AAAAABUTSALP2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMJWGMYTKMJTGA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

hagikent · 2025-01-27T19:33:48Z

@hanhou @XX-Yin
5_2025-01-27_11-06-05.json
behavior_session_model_5_2025-01-27_11-06-05.json
python_gui_log.txt
rig.json
session.json

XX-Yin and others added 24 commits January 3, 2025 17:12

adding second stimulus to the bonsai workflow

bb6d107

adding give second stimulus panel

c36ec07

adding GiveSecondStimulus and SecondStimulusDelay

2849a09

sending the second delay stimulus

0bb939c

changing the reward delay time

bc20511

typo and renaming

c1d4a72

typo

03fd88d

show camera recording

e935cb8

receiving and saving the second sound

55547dc

changing name and logic of soundcard_eventN

c592dda

typo

75516de

move position

5344d40

move to the audio

0771e05

continue

c5accbd

comment

2654843

changing the name

42c8b8d

adding delay

4c0344a

adjusting the reward consume time

eff727b

changing to float

50ea9df

adding no reward

07e6e55

making the reward delay as a distribution

e446ae6

saving the reward delay and the second stimulus delay

749aeb8

changing the ephys gui

828611b

removing a label

3b72c21

XX-Yin mentioned this pull request Jan 6, 2025

Adding CS+/CS- and Reward Delay to task #1263

Open

XX-Yin changed the title ~~Ading second stimulus~~ Adding second stimulus Jan 6, 2025

hagikent requested review from hagikent and micahwoodard January 9, 2025 04:22

hagikent reviewed Jan 9, 2025

View reviewed changes

XX-Yin added 3 commits January 9, 2025 12:43

adding take one

60ae706

changing the name RewardDelay to SecondStimulusDelay

cf50fe8

removing testing line

f40567b

XX-Yin added 6 commits January 10, 2025 16:39

adding independent randomness

d000a98

adding independent randomness

7c58b6d

changing the name

e0db7d8

moving the position

59271fb

updating the ephys gui

3d5e245

set default as even

0b1366d

		@@ -392,16 +395,23 @@ def _generate_next_trial_other_paras(self):
		# get the ITI time and delay time
		if self.TP_Randomness=='Exponential':

Adding second stimulus #1276

Are you sure you want to change the base?

Adding second stimulus #1276

Uh oh!

Conversation

XX-Yin commented Jan 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request instructions:

Describe changes:

What issues or discussions does this update address?

Describe the expected change in behavior from the perspective of the experimenter

Describe any manual update steps for task computers

Was this update tested in 446/447?

Uh oh!

hagikent commented Jan 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bruno-f-cruz commented Jan 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

XX-Yin commented Jan 7, 2025

Uh oh!

rachelstephlee commented Jan 7, 2025

Uh oh!

rachelstephlee commented Jan 8, 2025

Uh oh!

XX-Yin commented Jan 8, 2025

Uh oh!

hagikent commented Jan 8, 2025

Uh oh!

XX-Yin commented Jan 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

hagikent commented Jan 9, 2025

Uh oh!

hagikent left a comment

Choose a reason for hiding this comment

Uh oh!

hagikent Jan 9, 2025

Choose a reason for hiding this comment

Uh oh!

rachelstephlee Jan 27, 2025

Choose a reason for hiding this comment

Uh oh!

XX-Yin commented Jan 9, 2025

Uh oh!

hagikent commented Jan 9, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

XX-Yin commented Jan 9, 2025

Uh oh!

hagikent commented Jan 9, 2025

Uh oh!

XX-Yin commented Jan 9, 2025

Uh oh!

XX-Yin commented Jan 13, 2025

Uh oh!

rachelstephlee commented Jan 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

hagikent commented Jan 27, 2025

Uh oh!

hanhou commented Jan 27, 2025 • edited by hagikent Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rachelstephlee commented Jan 27, 2025 via email

Uh oh!

hagikent commented Jan 27, 2025

Uh oh!

Uh oh!

XX-Yin commented Jan 5, 2025 •

edited

Loading

hagikent commented Jan 6, 2025 •

edited

Loading

bruno-f-cruz commented Jan 7, 2025 •

edited

Loading

XX-Yin commented Jan 8, 2025 •

edited

Loading

hagikent commented Jan 9, 2025 •

edited

Loading

rachelstephlee commented Jan 27, 2025 •

edited

Loading

hanhou commented Jan 27, 2025 •

edited by hagikent

Loading