Summary: Our ability to process sentences relies on the dynamic nature of working memory, where information is stored and integrated with our future intentions.
New research reveals that visual memories adapt according to our future use of that information. These findings challenge conventional theories arguing that our working memory’s neural codes remain unchanged over time.
Instead, the study reveals that our brains dynamically reformat these memories to better align with potential future actions based on these recollections.
Key Facts:
- Working memory doesn’t just store information; it adapts it based on how we intend to use it in the future.
- Unlike traditional beliefs that our memory codes remain stable, recent studies show these codes change dynamically over time.
- The neural dynamics of working memory aim to reformat memories to correlate with our future actions based on those memories.
Source: NYU
When we use our working memory, we temporarily retain information in our brain. For instance, you are able to comprehend this sentence because you are briefly storing in working memory each of the words you are reading until you put them together to form the meaning of the sentence.
The importance of working memory to many of our cognitive abilities is well known, but less clear are the neurological machinations driving this process.
A team of researchers has now demonstrated that the key to understanding working memory relies not only on what one is storing in memory, but also why. This is the “working” part of working memory, which emphasizes the purpose of storing something in the first place.
“We now know that our visual memories are not simply what one has just seen, but, instead, are the result of the neural codes dynamically evolving to incorporate how you intend to use that information in the future,” explains Clayton Curtis, a professor of psychology and neural science at New York University and the senior author of the paper, which appears in the journal Current Biology.
Specifically, the study focuses on both how we store the visual properties of our memories in the occipital lobe, where our visual system resides, and on how the neural codes that store those memories change over time as people begin to prepare a response that depends on the memory.
In the Current Biology study, the response simply required people to look where they remembered an object that disappeared several seconds ago.
“The research makes it clear that memory codes can simultaneously contain information about what we remember seeing and about the future behavior that depends on those visual memories,” notes Hsin-Hung Li, an NYU postdoctoral researcher and the paper’s lead author.
“This means the neural dynamics driving our working memory result from reformatting memories into forms that are closer to later behaviors that rely on visual memories.”
Textbook theories state that the storage codes for our working memory are stable over time. This means that the pattern of neural activity that stores a given visual memory is the same as when it was first seen and encoded—whether it is a second later or 10 seconds later.. These patterns of neural activity store visual memories, providing a bridge across time between a past stimulus and a future memory guided response.
However, recent studies of animals indicate that these neural patterns are much more dynamic—in fact, the memory codes are not stable and, instead, appear to change over time in puzzling ways.
To explore this, Li and Curtis, who previously uncovered how our working memory is formatted in the brain, devised innovative methods to both measure changing neural dynamics and critically make the dynamics interpretable.
To do so, they projected the complex neural measurements into a simple 2D plane, like the screen of your laptop or smartphone.
The video below depicts how the pattern of neural activity evolves during a working memory trial. Initially, you can see a bump of activity encoding the briefly presented visual target (pink circle) in both primary visual cortex (V1) and a high-level visual area (V3AB).
In V3AB, this bump remains at the target location throughout the memory delay. However in V1, a line of activity evolves during the delay between where the person is looking (pink cross) and where they will move their eyes after the delay.
The researchers believe that this line reflects the trajectory of the shift of gaze that is being rehearsed in people’s minds, but has yet to be executed.
Although previous work had documented neural dynamics during working memory, the reason for why these dynamics occur had remained unknown.
These new results help address this mystery. They indicate that the dynamics reflect transformations of past sensory events— what we have just seen—into future memory guided behaviors—what we might do with the memory.
“We’ve now shown that mnemonic codes can simultaneously contain information about a past remembered stimulus and the subsequent behavior that depends on that stimulus,” observes Curtis. “The neural dynamics of our working memory result from reformatting memories in order to align them with how we use them in the future.”
Funding: The research was supported by National Institutes of Health grants from the National Eye Institute [NEI] (R01 EY-016407, R01 EY-027925, and R01 EY-033925). Hsin-Hung Li was supported by the Swartz Foundation Postdoctoral Fellowship.
About this visual neuroscience and memory research news
Author: James Devitt
Source: NYU
Contact: James Devitt – NYU
Image: The image is credited to Neuroscience News
Original Research: Closed access.
“Neural population dynamics of human working memory” by Clayton Curtis et al. Current Biology
Abstract
Neural population dynamics of human working memory
Highlights
- Both stable and dynamic neural codes support visual spatial working memory (WM)
- Surprisingly, WM dynamics are greater in visual compared to frontoparietal cortex
- Neural dynamics were made interpretable by modeling population-level activity
- Reformatting of WM representations drives neural dynamics
Summary
The activity of neurons in macaque prefrontal cortex (PFC) persists during working memory (WM) delays, providing a mechanism for memory.
Although theory, including formal network models, assumes that WM codes are stable over time, PFC neurons exhibit dynamics inconsistent with these assumptions.
Recently, multivariate reanalyses revealed the coexistence of both stable and dynamic WM codes in macaque PFC.
Human EEG studies also suggest that WM might contain dynamics. Nonetheless, how WM dynamics vary across the cortical hierarchy and which factors drive dynamics remain unknown. To elucidate WM dynamics in humans, we decoded WM content from fMRI responses across multiple cortical visual field maps.
We found coexisting stable and dynamic neural representations of WM during a memory-guided saccade task.
Geometric analyses of neural subspaces revealed that early visual cortex exhibited stronger dynamics than high-level visual and frontoparietal cortex. Leveraging models of population receptive fields, we visualized and made the neural dynamics interpretable.
We found that during WM delays, V1 population initially encoded a narrowly tuned bump of activation centered on the peripheral memory target.
Remarkably, this bump then spread inward toward foveal locations, forming a vector along the trajectory of the forthcoming memory-guided saccade. In other words, the neural code transformed into an abstraction of the stimulus more proximal to memory-guided behavior.
Therefore, theories of WM must consider both sensory features and their task-relevant abstractions because changes in the format of memoranda naturally drive neural dynamics.
