C222 - Githubissues

Brody-Lab / jbreda_animal_training

Repository for the ingestion, cleaning and visualization of behavioral experiments

0 stars 0 forks source link

C222 #30

Open jess-breda opened 9 months ago

jess-breda commented 9 months ago

turning up hit switch threshold to 0.5 now that animal is starting to grasp light guidance

jess-breda commented 9 months ago

pro performance is starting to drop- increasing the hit threshold from 0.5 -> 0.75

jess-breda commented 8 months ago

putting back to pro only due to rig swtich

jess-breda commented 8 months ago

going to try the 1/2 XOR blocks with this animal

switch sound on
hit switch at 0.7

jess-breda commented 8 months ago

doing well with the 1/2 XOR and light guide. Delaying anti with alpha minus at 0.68 since most animals are just staying in place at 0.75

jess-breda commented 7 months ago

alpha minus dropped from 0.68 -> 0.64

jess-breda commented 7 months ago

rig switch- seems to be doing well with the sound-light task, but give use isn't dropping yet.

jess-breda commented 7 months ago

got off guide for 1/2 XOR. put into next stage (no guide) with smaller block size (15 -> 20) and slightly higher hit threshold (0.7 -> 0.72) and switch sound 5 -> 2 seconds

jess-breda commented 6 months ago

performance for this animal is looking really good. Going to turn off the switch sound and see how performance changes. Blocks are currently static. Min size is being set 20 -> 15.

jess-breda commented 6 months ago

putting in interleaved with max blocks at 7

jess-breda commented 6 months ago

seems to be getting interleaved!

jess-breda commented 6 months ago

max blocks 7 -> 5

jess-breda commented 6 months ago

taking this approach to seeing how this animal does with this approach and implemented it today

jess-breda commented 6 months ago

day 2 of adding in 1/2 XOR 4th pair. This animal is doing pretty well! Accidentally went into random due to max blocks being at 5.

Want to make switching threshold a bit higher since maintaining pro doesn't seem to be an issue so made the following changes:

switch threshold 0.72 -> 0.76
sampled block size mean 15 -> 18
max blocks 5 -> 11

jess-breda commented 6 months ago

When plotting by stim pair- this animal has resorted to an sb only rule. This makes sense as it get's them to perform over chance. I think their 1/2XOR wasn't solid enough. I want to make sure they can do 1/2XOR form the start of the sessions so I am putting them back into that to make it more robust

jess-breda commented 6 months ago

Decent recovery- we will see how it goes

jess-breda commented 6 months ago

turning max blocks 9 -> 7 to start getting some random

jess-breda commented 6 months ago

really starting to get it again- max blocks 7 -> 5

jess-breda commented 6 months ago

max blocks 5 -> 3

jess-breda commented 5 months ago

seems to have gotten it much better, putting pro prob 0.6 -> 0.5

jess-breda commented 5 months ago

put max blocks to 1 to see how well this animal does fully interleaved