2022
DOI: 10.1038/s41593-021-01007-z
|View full text |Cite
|
Sign up to set email alerts
|

Mice alternate between discrete strategies during perceptual decision-making

Abstract: Classical models of perceptual decision-making assume that animals use a single, consistent strategy to form decisions, or that decision-making strategies evolve slowly over time. Here we present new analyses suggesting that this common view is incorrect. We analyzed data from two mouse decision-making experiments and found that choice behavior relies on an interplay between multiple interleaved strategies. These strategies, characterized by states in a hidden Markov model, persist for tens to hundreds of tria… Show more

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

11
219
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
5
1
1

Relationship

1
6

Authors

Journals

citations
Cited by 148 publications
(230 citation statements)
references
References 74 publications
11
219
0
Order By: Relevance
“…Although we might expect rodents to be able to quickly figure out this task and become fully committed to the inference-based strategy, this was not the case. Instead, the frequent switches between behavioral states is representative of rodent behavior and agrees with many other studies of a diverse array of tasks 27,28 . This feature of rodent behavior once again highlights the need for powerful analytical methods that can infer hidden behavioral states that govern behavior, since these types of models allow a finer scale resolution when dissecting the behavioral circuits.…”
Section: Discussionsupporting
confidence: 88%
See 1 more Smart Citation
“…Although we might expect rodents to be able to quickly figure out this task and become fully committed to the inference-based strategy, this was not the case. Instead, the frequent switches between behavioral states is representative of rodent behavior and agrees with many other studies of a diverse array of tasks 27,28 . This feature of rodent behavior once again highlights the need for powerful analytical methods that can infer hidden behavioral states that govern behavior, since these types of models allow a finer scale resolution when dissecting the behavioral circuits.…”
Section: Discussionsupporting
confidence: 88%
“…2e-g), our model takes inspiration from recent modeling approaches which are used to infer discrete latent states that underlie neural dynamics 37 , natural behavior 38 , and behavior in decision-making tasks 27,28 . In particular, we adapted the recent GLM-HMM framework 28 , where discrete hidden states determine the coefficients of a generalized linear model (GLM) which specifies how the decision of the animal depends on external trial variables. While the latent states in this approach are updated from trial to trial, latent states in the blockHMM framework govern the choice selection across entire blocks, and are only updated at the boundaries of block transitions.…”
Section: Discussionmentioning
confidence: 99%
“…As a result, state transitions and observations can be conditioned on the input at the current time step. Also known as GLM-HMMs in the neuroscience literature [2,9,10,19], these models allow transitions and observations to be generated by state-dependent generalized linear models (GLMs).…”
Section: Input-output Hidden Markov Modelsmentioning
confidence: 99%
“…4B. We set the parameters to match those in [2], which fit the IO-HMM to the binary choice data of mice performing a decision-making experiment. Each GLM has a weight associated with the external stimulus (w k ) presented to the mouse, as well as a bias parameter (b k ), such that w k = {w k , b k }.…”
Section: Experiments With Io-hmmsmentioning
confidence: 99%
See 1 more Smart Citation