summary: When it comes to decision making, rather than sticking to a single strategy, the brain computes multiple decision strategies simultaneously.
source: Champalimaud Center for the Unknown
Choosing a line at the supermarket checkout may seem like a no-brainer, but it may actually involve a complex series of brain math operations. Perhaps you count the number of shoppers per line and choose the shortest, or estimate the number of items on each conveyor belt. Perhaps quickly weighing shoppers and objects and perhaps even the apparent speed of a cashier… In fact, there are many strategies to solve this problem.
So how does the brain know how to make decisions in such situations where there are many possible strategies to choose from?
The study was published today, April 13 in the journal Natural neuroscience He provides a surprising answer to this question by showing that rather than sticking to a single strategy, the brain can compute multiple alternative decision strategies simultaneously.
The study, which was led by Fanny Cassets and senior authors Zachary Menin and Alfonso Reynart, at the Champalimaud Foundation in Lisbon, Portugal, conducted a custom-designed experiment that used a type of “virtual reality” setting for mice, in which the animals were tasked with searching for water in a virtual world.
Specifically, the authors designed a “mouse virtual world” that contained the kind of foraging problem that animals’ brains evolved to be good at solving, allowing them to study the complex decision-making strategies used by mice.
Any place in the virtual world can provide water unreliably and, at some point, “dry up” and stop providing water altogether. The rats had to decide when to leave one location and move to another in search of more water.
To solve the task optimally, the best strategy would be for the rats to learn to count the number of consecutive missed attempts to obtain water at a given location, and to switch locations when the number of consecutive failures is large enough.
But there were multiple alternative strategies for addressing the series of successful and unsuccessful attempts, including, for example, calculating the difference between the number of successful and unsuccessful attempts.
Each strategy combines errors and successes across time in a certain way, and thus has a distinct time course—which is called a “decision variable”—that can be matched to the time course of patterns of brain activity.
The researchers recorded activity from large groups of individual brain cells in a part of the brain known as the motor cortex while the mice were performing the task. They then looked for groups of temporal profiles of recorded premotor neuron activity that resembled decision variables associated with different strategies.
To the authors’ surprise, the data showed that while each mouse focused on its own strategy, their brains did not.
“We found that while activity in the premotor cortex reflects the computation the mouse was actually using, it also reflects alternative decision variables useful for the same task, and even decision variables useful for other tasks,” explains Fanny Cassets.
Zach Minnen, one of the study’s senior authors, adds that, “Unlike our experiment with checkout lines, we found that the brain can actually execute several different counting strategies at the same time, which is reminiscent of the concept of superposition in quantum mechanics.”
Although there is still much to be explored in this area, this study provides an important basis for future research.
Our findings indicate the need for new ways of thinking about the basic processes involved in decision-making and action selection. One of our next steps will be to investigate how the brain chooses between different decision variables and how these decisions are translated into action,” says Fanny Cassets.
What could be the benefit of representing both used and unused strategies simultaneously?
“This arrangement may facilitate cognitive flexibility and learning, because changing strategies requires only attention to the prior valid decision variable, rather than having to build it from scratch,” says Alfonso Renart, another senior author.
These findings have important implications for our understanding of how the brain processes and selects decision variables in complex environments. There may be implications for developing more flexible and adaptable machine learning systems, which may be particularly useful in situations where there is a high degree of uncertainty or complexity,” concludes Zach Menen.
About this Neuroscience Research News
author: Anna Gershenfeld
source: Champalimaud Center for the Unknown
communication: Anna Gershenfeld – Champalimaud Center for the Unknown
picture: The image is in the public domain
Original search: Closed access.
“Reservoir of Foraging Decision Variables in the Mouse Brain” by Fanny Cazettes et al. Natural neuroscience
Reservoir of foraging decision variables in the mouse brain
In any given situation, the environment can be analyzed in different ways to produce decision variables (DVs) that determine which strategies are useful for different tasks. It is generally assumed that the brain only computes a single DV that determines the current behavioral strategy.
Here to test this assumption, we recorded neural groups in the frontal cortex of mice performing a foraging task while accepting several DVs. Methods developed to detect DV currently in use revealed the use of multiple strategies and occasional switches in the strategy during the sessions.
Optogenetic manipulations showed that the secondary motor cortex (M2) is necessary for mice to use different DVs in a task. Surprisingly, we found that regardless of which DV best explains the current behavior, M2 activity simultaneously encodes a whole core set of computations that select a stock of DVs suitable for alternative tasks.
This form of neuronal multiplexing may confer significant advantages for learning and adaptive behaviour.