Arithmetic on GlobalCounters for State transition

Vlad_NAK

Dear Josh,

I am developing a simple auditory detection task for head restrained mice.
The reward should only be delivered only if the mouse has licked more after the sensory stimulation (LickCountResponse) than before (LickCountBaseline).
Ideally, I would like to use the following condition to trigger the reward delivery state: LickCountResponse>(2*LickCountBaseline) .
I am collecting the lick counts thanks to the GlobalCounters but I can not implement this ratio as StateChangeCondition within my state machine.
Based on the wiki and the content of the forum, it looks like it is not feasible ... Am I right?

My current solution is to split a single trial into 2 state machines that run sequentially, it is not perfect, but it works:

1) Run a first state machine to trigger the sensory stimulation and collect both LickCountBaseline and LickCountResponse
2) Calculate the lick ratio in Matlab
3) Run a second state machine that either delivers a reward or not, based on the lick ratio calculated in Matlab
It works but with little problems:
I am blind during 2) regarding datacollection, which should be negligible (it is inconsistent but only lasts approximately from 5ms to 8ms only)
I would like to keep the PWM1 ON during the whole trial, but it switches OFF at step 2) of the trial when no state machine is running ... and unfortunately the short light blink is visible and could be interpretated as a visual cue by the mouse

Do you think there is a more efficient way to achieve this trial structure? Maybe with a single state machine run?
Can I use Matlab to force the PWM1 to be ON when no state machine is running ?

Thank you in advance for your time and expertise

Josh

Hello Vlad,

This is one task that's a bit awkward to stuff into state machine formalism!
If your lick counts are relatively low (e.g. under 64 licks in the baseline period) you can build the trial as:

Add state 1 to trigger a global timer that drives the sensory stimulation via its associated channel (e.g. this). The 'OnsetDelay' should equal the baseline period. The 'OnsetTrigger' property should trigger a second global timer, whose 'Duration' is the response period.
Baseline period (includes state 1): Add states 2 - 64. Each lick progresses through the series of states, and each state moves to a Response Period state (see point 3) in response to the 'GlobalTimer1_Start' event (coincident with the stimulus).
Response period: Add states 66-194. Each lick progresses through the series of states. A lick in state 194 moves to the reward state. In each state, 'GlobalTimer2_End' exits the trial with no reward.

You'd map states in the baseline period to states in the response period proportional to the number of response licks necessary for reward, e.g. state 4 --> state 186, since 8 licks would be necessary to move from there to reward.

Of course this setup is awkward and puts a cap on the number of licks, but it's an option that doesn't require two trials.

If you end up with two trials, you may be able to cut out the dead time with TrialManager.

I hope this helps,
Josh

Vlad_NAK

Hello Josh and thank you a lot for your answer,

I will try to implement the strategy you propose and also try to be more familiar with the TrialManager that will likely facilitate my future akward tasks 🙂

A last quick question is about which Matlab command can I use to force the PWM1 to be ON when no state machine is running (similarly to what is done when clicking on LED boolean of the Bpod GUI Console) ?

I tried to set the PWM1 from 0 to 255 changing its corresponding value in BpodSystem.HardwareState.OutputOverride or OutputState, but it failed. How can I set it to remain ON without running a state machine ?

Cheers,

Vlad

Edit:
Just played with the TrialManager that seems convenient for my purpose, making the dead time between my two StateMachine negligible :

trialManager=BpodTrialManager; SendStateMachine(StateMachine1) trialManager.startTrial(); RawEventsTriggerState= trialManager.getCurrentEvents({'TriggerState'});

I added a 'Buffer' state named TriggerState to my StateMachine1 which allows the extraction of licks when this state is reached. I then compute the lick ratio on Matlab and send the desired following StateMachine2 while StateMachine1 still run :

if LickCountResponse>(2*LickCountBaseline) SendStateMachine(StateMachine2Reward,'RunASAP') else SendStateMachine(StateMachine2NoReward,'RunASAP') end

Thanks for your advices 😉

Josh

No problem, glad to help!