by Summary: The brain creates specific and distinct spaces within the cortex for each general rule of working memory and controls those spaces with brain rhythms, researchers report.
Source: WITH
Routine tasks that require working memory, like baking, involve memorizing both some general rules (e.g. read the oven temperature and time from the recipe and then put it on the oven) and some specific content for each instance ( e.g., 350 degrees for 45 minutes for a loaf of rye, but 325 degrees for 8 minutes for biscuits).
A new study provides a novel explanation for how the brain uniquely manages the general and specific components of such cognitive demands.
The research, led by scientists from MIT’s Picower Institute for Learning and Memory and the Karolinksa Institute and KTH Royal Institute of Technology in Stockholm, Sweden, shows that for each general rule, the brain creates different spaces in the cortex and controls these patches with brain rhythms , a concept the authors call “spatial computing”.
This system, evident in the study’s animal experiments, explains how the brain can easily maintain a consistent understanding of a process even when the specific contents are constantly changing (like the time and temperature for bread or biscuits).
It also answers some questions that neuroscientists have struggled with about the physiological processes underlying working memory.
“Your brain can generalize instantly. If I teach you to follow some rules like remembering C, A, and B and place them in alphabetical order, and then I switch the content to F, D, and E, you won’t miss a beat,” said Earl K Miller, Picower Professor at MIT’s Picower Institute for Learning and Memory and co-author of the study in nature communication.
“Your brain can do this because it represents the rules and content on different physical scales. One can simply be plugged into the other.”
functioning of working memory
Years of research in Miller’s lab, largely led by lead author Mikael Lundqvist, who now works at Karolinska, has shown that working memory tasks are driven by an interplay of brain rhythms at different frequencies. Slower beta waves carry information about task rules, selectively giving way to faster gamma waves when it comes time to perform operations such as storing information from the senses or reading out information when retrieval is required.
But these waves operate on networks of millions of neurons, only a fraction of which store the individual pieces of information that are relevant at any given moment. In addition, neurons containing information about specific objects can be found everywhere. Some become electrically excited, or “spike,” in response to different task rules than others, and they often tend to spake at least somewhat even when their information is not relevant.
So how can these rather imprecise rhythms selectively control just the right neurons at the right time to do the right things? Why are neurons spiking on specific elements scattered and redundant? What makes one neuron designed specifically for “350 degrees” perk up when that information needs to be stored, but make another neuron perk up with that information when it needs to be retrieved?
Researchers realized that all of these questions could be solved by spatial computing theory. Individual neurons that represent pieces of information can be widely scattered across the cortex, but the rule that applies to them is based on the patch of the network in which they reside. These patches are determined by the pattern of beta and gamma waves.
“By analyzing many individual neurons over the years, we have always wondered why so many of them appear to behave similarly,” Lundqvist said.
“Regardless of whether they preferred the same external stimulus or not, many neurons showed similar patterns of activity during working memory. And these patterns changed from task to task. It also appeared that neurons that were closer together within the prefrontal cortex exhibited the same pattern more often. We thought that memory representations might actually dynamically flow around the prefrontal cortex to implement task rules.”
Suppose your friend calls you at the gym and asks you to pick up a watch that he accidentally left in his locker. To do this, the padlock wheels must be turned to the numbers in the combination (e.g. 24, 17, 32). Spatial computing says that when you hear the combination, your brain creates different patches for each step (first, second, third).
In each patch, the neurons representing the combination number of that particular step are particularly excited by gamma waves applied at the time the rule is relevant (i.e. 24 in the “first” patch, 17 in the “second” patch and 32 in the “third” patch).
In this way, individual neurons that encode specific information can be selectively assigned general rules by having the brainwaves control the areas in which they are located. In any given patch, all neurons may be somewhat excited by the gamma waves, but those representing the element that conforms to the rule will spike the most.
“In this way, memory representations could be dynamically reshaped to meet current task demands, regardless of how individual neurons are connected or what stimulus they prefer,” said co-senior author Pawel Herman of KTH. “It could explain our impressive generalization abilities in novel situations.”
That’s not to say that every patch is fixed forever. The patches can come and go as needed, wherever the brain shapes them for the task at hand. There is no permanent “remember oven temperature” patch in the brain.
“It gives the brain flexibility,” Miller said. “Kognition is all about flexibility.”
Experimental Evidence
The researchers didn’t just come up with theories. To test spatial computing in real physical brains, they made four experimental predictions about what to observe when animals played working memory games, such as B. Remembering a series of images in a sequence.
The first prediction was that there should be clear neural signals about the rules and information about individual items. In fact, the team measured that wave outbursts carried regulatory information. Individual neural spikes, meanwhile, carried a mix of individual items and task rules consistent with representing individual items and having specific rules imposed on them.
The second prediction was that rule information should be spatially organized, and the third prediction was that these rule-enforcing spatial patterns should be consistent as long as the rules of the game remained the same, regardless of whether the individual elements changed.
In fact, the researchers found that there were different locations for gamma-ray bursts for different rules, and that these remained stable even when individual items varied during each game.
The final prediction was that brainwave activity should cause neural spiking activity to present the right information at the right time. This was also reflected in the experimental observations.
The researchers saw different brainwave patterns for when the brain needed to store images in memory and when it needed to recall the “right” one. In general, beta waves were more reduced and neurons were more heavily and widely spiked during retrieval than during storage.
The paper does not answer every question about working memory. It’s not yet clear how neurons encoding specific information in one patch might associate with their brethren in another patch, or how the brain controls the patches. More research can answer these further questions about the implications of the new theory of spatial computing.
About this news from memory and neuroscience research
Author: press office
Source: WITH
Contact: Press Office – MIT
Picture: The image is in the public domain
Original research: Open access.
“The dynamics of working memory control follow the principles of spatial computing” by Earl K. Miller et al. nature communication
Abstract
The control dynamics of working memory follow the principles of spatial computing
Working memory (WM) allows us to remember and selectively control a limited set of items. Neural evidence suggests that this is accomplished through interactions between bursts of beta and gamma oscillations. However, it is not clear how oscillations reflecting the coherent activity of millions of neurons can selectively control individual WM elements.
Here we propose the novel concept of spatial computing, in which beta and gamma interactions cause object-specific activities to flow spatially across the network during a task.
In this way, control-related information such. B. the order of items, stored in the spatial activity, independent of the detailed recurring connectivity that supports the item-specific activity itself.
The spatial flow is in turn reflected in lower dimensional activity shared by many neurons. We verify these predictions by analyzing local field potentials and neuronal spiking.
We hypothesize that spatial computing can facilitate generalization and zero-shot learning by using the spatial component as an additional information-encoding dimension.
