Memories learned within a distorted coordinated system are also distorted when recalled later


In order to remember where important events happened, or how to get from A to B, our brains form mental “maps” of our environment.

An important component of these mental maps are the so-called grid cells.

Different grid cells are active when we occupy different locations in an environment, creating a characteristic pattern of activity.

This pattern consists of equilateral triangles that form a symmetrical grid structure.

The discovery of grid cells, in the brains of rats, was awarded the Nobel Prize in 2014. Scientists suspect the same is true in human brains.

Our mental maps and grid patterns formed from grid cells allow us to remember where a certain place is located and determine how far away it is from other locations. This should all work well if the grid patterns are symmetrical and regular.

However, if the patterns are disturbed, our mental maps may become inaccurate. A team of researchers from the Max Planck Institute for Cognitive and Neurosciences in Leipzig, University College London, and the Kavli Institute for Systems Neurosciences in Norway have pursued this idea.

In an earlier experiment, by English neuroscientists, the activity of grid cells in rats was recorded while they made their way through different enclosures.

It became clear that under certain circumstances the grid cells lost their hexagonal symmetry and “fired” more irregularly.

If the animal navigated through a square box, the perfect grid pattern in the rat brain could be detected.

However, if it moved through a trapezoidal enclosure, the grid pattern was far less regular.

The coordinate system of the grid cells in rats seemed to be distorted under these circumstances. Could this have consequences for the accuracy of our mental maps?

“It should lead to distortions in our memory, if our brain really uses this coordinate system,” thought Christian Doeller and colleagues.

“We have therefore carried out an experiment, with virtual reality, in which the test subjects learn different positions in space.

They first do this in a square environment, where the coordinate system should work well, and then in a trapezoidal environment, where the coordinate system of the grid cells should get distorted”, explains Jacob Bellmund.

The participants wore virtual reality glasses and navigated through the virtual environments using a 360° motion platform.

Each environment contained six objects and they learned which object belonged at which position in the environment. The platform provided a realistic running sensation, with their feet gliding over it in a kind of “moonwalk” (see video).

In actuality, the participants only got off the ground in the virtual world.

“We then compared how exactly the participants were able to learn the positions. As expected, they were worse in the trapezoidal environment than in the square one.

In the trapezoidal environment, they were especially bad in the narrow half. This would correspond exactly to the area with the biggest distortions of the grid cell coordinate system,” explains Bellmund.

This shows a grid map

What happens if the coordinate system of our brain, which measures our mental maps, is distorted? The image is credited to MPI CBS.

The scientists then wanted to know whether these distortions would remain in memory, even if the participants were no longer in the asymmetrical environment.

Meaning, their mental coordinate system should be ‘square’ again. To do this, they asked participants to estimate the distance between pairs of objects.

Bellmund and his team had arranged the objects in such a way that the actual distance between the pairs was always identical.

But, if there were distortions in the participants’ memories the same distances should be recalled as shorter when recalling the trapezoid objects than those in the square.

Within the trapezoid, the distances were remembered as longer in the narrow half than in the wide half. Thus, memories that were learned within a distorted coordinate system are also distorted when remembered later.

It is precisely these distortions of our mental maps that we can predict with a model coordinate system,” says Jacob Bellmund.

Previous work by the MPI scientists has suggested that the brain not only creates mental maps to find its way, but that other cognitive processes are also overlaid on our brain’s navigational system.

Our memories do not exist in isolation, and neither do the neural circuits that represent them. Experiences may produce transient records in working memory—a temporary store for information to be maintained and manipulated over delays of seconds (Baddeley, 1992; Baddeley & Hitch, 1974; Repov & Baddeley, 2006).

Experiences can also simultaneously lay down more lasting traces as episodic memories, available to be recalled at a later time (beyond minutes), allowing us to relive specific, previously experienced events tied to the time and place of their occurrence (Tulving, 1983).

Early models proposed that working memory and long-term memory operated wholly in parallel (Shallice & Warrington, 1970). Evidence for the dissociation between working memory and episodic memory largely came from lesion studies, which found that damage to the medial temporal lobe (MTL) caused severe episodic memory deficits (Cave & Squire, 1992; Squire, 1992), while working memory, associated with the prefrontal cortex (Cohen et al., 1994), remained intact (Drachman & Arbit, 1966). More recent models propose that they support each other (Baddeley & Hitch, 2000; Cohen & O’Reilly, 1996).

There is accumulating evidence that episodic memory, and its neural substrates in the MTL, are engaged during short-term memory tasks that also engage working memory (Axmacher et al., 2007; Lewis-Peacock, Cohen, & Norman, 2016; Ranganath, 2005; Ranganath & Blumenfeld 2005; Ranganath, Cohen, Dam, & D’Esposito, 2004; Ranganath, D’Esposito, Friederici, & Ungerleider, 2005), suggesting these memory systems do not operate entirely independently of one another.

Experiments testing for an interaction between episodic memory (EM) and working memory (WM) have historically focused on the hypothesis that EM is used to support WM when maintenance is disrupted, leading to errors that reflect features of EM. For instance, participants show proactive interference from recently studied stimuli when WM is disrupted for 18 seconds (Wickens, Dalezman, & Eggemeier, 1976).

However, subsequent research suggests that EM may contribute to WM more ubiquitously, even when WM is not disrupted during 4-second delays (Atkins & Reuter-Lorenz, 20082011). Here, we investigate the nature of these interactions to ask how EM contributes to undisrupted WM.

Do ongoing reinstatements from episodic memory influence working memory, even in the absence of distraction?

A growing number of studies indicate that during periods of rest, the neural structures that support EM are active (Buckner, 2010) and appear to be reinstating recent experiences (Tambini, Ketz, & Davachi, 2010) or activating potential future scenarios constructed on the basis of past experiences (Buckner & Carroll, 2007).

These reinstatements trigger coordinated activity patterns across a broad swath of cortical regions, including those presumably involved in WM maintenance, such as the prefrontal cortex (Miller & Cohen, 2001).

This widespread activation is reliably present even during brief lapses in external stimulation (Logothetis et al., 2012), such as those typically used as maintenance periods in WM experiments.

These observations lead us to ask the question: How do ongoing reinstatements from EM affect the content of WM, even when the latter is not being disrupted? We hypothesized that an influence of EM on WM search might be observable by more sensitive measures than substitution errors during recall: through examination of reaction times (Atkins & Reuter-Lorenz, 20082011) and the use of content-specific pattern analysis in neuroimaging.

Using context as a signature of episodic memory

To test our hypothesis, we leverage the fact that retrievals from EM carry with them temporal and associative context (Howard & Kahana, 2002), such that triggering the recall of one memory from a given context can cause the subsequent, involuntary recall of other memories sharing that context (Bornstein & Norman, 2017; Hupbach, Gomez, & Nadel, 2009). This can occur even at the short delays typically associated with WM (Hannula, Tranel, & Cohen, 2006).

Therefore, we reasoned that if reinstatements from EM occurred during WM maintenance, then these reinstatements would likely be of memories that shared an encoding context with the target stimuli.

Even if these reinstated memories do not lead to overt errors, they may intrude on or degrade other, task-relevant representations being maintained in WM, and thereby affect search and response times on subsequent decisions—even several seconds later, and even in the absence of further EM reinstatement (Atkins & Reuter-Lorenz, 2008).

They may also express themselves in patterns of neural activity reflective of the reinstated memories.

It is also possible that episodic memories are reinstated at the moment of retrieval instead of or in addition to during WM maintenance. Research on prospective memory, a memory task in which an individual must remember to perform an action at a target event in the future (e.g., remembering to stop at the supermarket on the way home; see Brandimonte, Einstein, & McDaniel, 1996), point to a reason why EM reinstatements only at probe could be strategic. Constantly monitoring the environment for the target event is cognitively costly; relying on environmental context clues to reinstate the intended action at the relevant decision point (e.g., getting into the car after work) could free cognitive resources for other tasks during the delay (McDaniel & Einstein, 2000).

Measuring the timing of memory reinstatements using neuroimaging over the course of a task can help address whether EM context reinstatements are ongoing or locked to retrieval.

Max Planck Institute
Media Contacts:
Jacob L.S. Bellmund – Max Planck Institute
Image Source:
The image is credited to MPI CBS.

Original Research: Closed access
“Deforming the metric of cognitive maps distorts memory”. Jacob L. S. Bellmund, William de Cothi, Tom A. Ruiter, Matthias Nau, Caswell Barry & Christian F. Doeller.
Nature Human Behaviour doi:10.1038/s41562-019-0767-3.


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.