RT @allemanjm: We live with the limitations of our memory, but don’t really know where they come from. Our new paper (.

7/7: Very excited about this paper! Culmination of a very long-term project led by @tafazolisina in collaboration with @neuro_flora, @adelardalan, @Nikola_T_Markov, @Mu_neuro, @marcelomattar and @nathanieldaw .

Join millions who have switched to Grok.

6/7: Finally, monkeys had to discover the task in effect, updating their internal belief based on feedback. As they learned which task was in effect, we found the task-relevant shared subspaces were gradually engaged and task-irrelevant information was compressed.

5/7: When performing a task, information was dynamically transformed from the relevant shared category subspace into the appropriate motor subspace. Suggests prefrontal cortex is a ‘global workspace’, where information flexibly moves between subspaces to perform different tasks.

4/7: Neural recordings found the stimulus’ color and shape category, and the motor action, were represented in separate subspaces of neural activity. These subspaces were shared across tasks – one could ‘build’ a task from the subspaces of other tasks.

3/7: So, we trained animals to perform three tasks. Each task required categorizing a stimulus input, based on either its color or shape, and then indicating the category by making one of two different types of motor responses. Tasks shared categorization and response components.

2/7: In particular, we wanted to test the hypothesis that the brain can reuse simple task ‘components’ to compositionally build complex tasks. For example, once we learn to tell if a piece of fruit is ripe, then we can use this as a component of foraging, cooking, and eating.

1/7: Humans and animals are remarkably good at performing many different tasks. On any given morning, one might transition from driving to work, to making coffee, to checking email, etc. We wanted to understand how the brain can learn and flexibly switch between multiple tasks.

New preprint! @tafazolisina shows the brain builds complex tasks by compositionally combining simpler sub-task representations. By dynamically reusing neural subspaces for sensory inputs and motor actions, the brain can flexibly perform multiple tasks. 🧵

Research was done by a wonderful team of Camden MacDowell @CamdenMacDowell, Brandy Briones @brandy_briones, Michael Lenzi, and Morgan Gustison @MorGusto.

In sum, this supports a ‘sampling hypothesis’ that suggests differences in behavior are due to differences in the top-down control mechanisms that govern the moment-to-moment selection of specific interactions between brain regions.

Rather, individuals differed in how frequently different motifs occurred. Each animal’s ‘barcode’ of motif expression explained differences in behavior and functional connectivity. In addition, individuals varied in how well they adapted their motif expression to new environments

Unexpectedly, this was not due to differences in how neural activity flowed across regions – the same spatiotemporal ‘motifs’ in cortical dynamics were seen across all individuals. This suggests differences in functional connectivity do not solely reflect structural differences.

To study how behavior and cortical neural dynamics differ across individuals, we combined an in utero valproic acid exposure model with widefield calcium imaging in mice. As expected, functional connectivity between cortical regions correlated with animals’ behavioral phenotype.

Behavior exists on a spectrum across individuals. One hypothesis is that differences in behavior reflect differences in how information flows through the brain. This is often studied through the lens of anatomical and functional connectivity.

New @CurrentBiology paper with @CamdenMacDowell relating individual differences in behavior to neural dynamics! Individuals differently expressed ‘motifs’ of cortex-wide neural activity, which explained variability in functional connectivity and behavior🧵

Project was led by @cjahn_neuro in collaboration with @Nikola_T_Markov, Britney Morea, @nathanieldaw and @BecketEbs.

5/5: Finally, attention re-mapped sensory stimuli into a common ‘value’ space which generalized across templates. This could allow the brain to use the same decision-making circuitry to decide which stimulus to select, regardless of changes in the template or environment.

4/5: Trial-by-trial decoding found neural representation of the template were updated incrementally, shifting towards features that were rewarded. This suggests reward learning can be used to learn attentional templates or, more generally, learn to control cognition.

3/5: Recordings in prefrontal and parietal cortex found attentional templates were semantically structured: templates for perceptually similar stimuli had similar neural representations. Such structured geometry could allow for generalization to new templates.