Posterior Parietal Cortex Encodes Memory Engrams

Post by: Amanda McFarlan

What's the science?

The hippocampus is a neural structure that is famously implicated in the study of learning and memory, and is known to be required for the rapid encoding of memory. The neocortex, however, has been traditionally thought to store memories more slowly, over time. Recent studies have shown that the posterior parietal cortex (a neocortical region also implicated in memory) may also be capable of rapidly acquiring memory representations known as ‘engrams’. A memory engram (i.e. memory trace) is hypothesized to be the way in which memories are represented and stored in the brain. So far, it has been unclear whether the posterior parietal cortex can encode true memory engrams. This week in Science, Brodt and colleagues used brain imaging to investigate dynamic changes in neocortical structures (including the posterior parietal cortex) during learning and memory.

How did they do it?

The authors used functional magnetic resonance imaging (MRI) to measure dynamic changes in neocortical brain activity in healthy participants (male and female) during an ‘encoding and recall’ memory task. The participants were tested in two conditions: experimental (39 participants) and control (33 participants). The experimental condition consisted of two sessions, 13 hours apart. In the first session, participants received diffusion-weighted MRI scans (as a baseline to measure later changes in brain microstructures) followed by a functional MRI (fMRI) scan during which they performed an object-location association task. The object-location association task was made up of two parts: encoding and cued recall. During encoding, participants were shown pairs of cards that appeared in their own unique spot on an 8x5 grid. Each card had an image of an item belonging to one of three categories: fruits and vegetables, animals or inanimate objects. Pairs of cards were shown one after the other for 2 seconds each, followed by a brief delay before the next pair. During cued-recall, participants were shown the first card in a pair and had to indicate the location of the second card in the 8x5 grid. Importantly, participants went through 8 repetitions of these encoding-recall runs, 4 in each session. After a short break, the participants received a second set of diffusion-weighted MRI scans followed by two high-resolution T1- and T2-weighted scans (which show brain anatomy). The second session (13 hours after the first session) consisted of a third set of diffusion-weighted MRI scans and an additional task and fMRI session. The object-location association task in the second session started with an initial recall to determine the retention memory from the first session. The control condition was almost identical to the experimental condition, except that participants did not undergo the object-location association task and fMRI scan.   

What did they find?

The authors used whole-brain analyses to identify active brain areas during memory recall in the first and second sessions. They found an experience-dependent increase in brain activity in the bilateral precuneus (part of the posterior parietal cortex), the dorsal visual stream, the cerebellum, thalamus, and motor areas with subsequent repetitions of the object-location association task. Moreover, increased activation in the bilateral precuneus persisted after 12 hours and was positively correlated with memory performance. These findings suggest that the posterior parietal cortex, especially the precuneus, may encode memory engrams. To be considered an engram, however, there needs to be evidence of structural plasticity in the brain. The authors analyzed diffusion-weighted MRI scans to investigate whether there were changes in mean diffusivity, a measure of water diffusion in the brain that can indirectly reflect learning-dependent plasticity. They found evidence of microstructural changes, as indicated by a decrease in mean diffusivity, in bilateral precuneus and areas along the dorsal and ventral visual streams. These changes were compared to diffusion-weighted scans in the control group to determine whether they were learning-specific. Indeed, the authors found evidence of learning-specific changes in the left precuneus, the left middle occipital gyrus and left fusiform gyrus that persisted for 12 hours and were positively correlated with memory performance. Altogether, these findings suggest that the posterior parietal cortex, especially the precuneus, may be an important structure for acquiring and storing memory engrams.

amanda (1).png

What's the impact?

This is the first study to provide evidence of a rapid neocortical engram in the posterior parietal cortex using a combination of fMRI and diffusion imaging. The precuneus has been shown to exemplify all the criteria for a memory engram: functional responses corresponding to memory recall that persist over time, with evidence of local structural plasticity.


Brodt et al. Fast track to the neocortex: A memory engram in the posterior parietal cortex. Science (2018). Access to the original scientific publication here.

The COMT Polymorphism and Cortical Thickness in ADHD

Post by Sarah Hill

What's the science?

How genetics relate to abnormal brain structure and function is currently an open question for researchers studying attention-deficit/hyperactivity disorder (ADHD). Research has suggested that a mutation in the catechol-O-methyltransferase (COMT) gene, a gene mediating dopamine signalling in the frontal cortex throughout neurodevelopment, is associated with altered gray matter volume in children with ADHD. However, which specific anatomical features and brain regions are related to COMT-associated gray matter abnormalities have yet to be determined. This week in Cerebral Cortex, Jung and colleagues report that the COMT mutation is specifically associated with differences in cortical thickness and surface area between children with and without ADHD.

How did they do it?

The authors used machine-learning methods to investigate whether cortical thickness and surface area were altered in children with ADHD and COMT gene mutations. First, they gathered MRI images from 39 boys with ADHD and 34 typically developing boys, all between the ages of 7 and 15, along with working memory scores for each participant. Next, they preprocessed the brain imaging data to acquire information on cortical thickness, surface area, and volume across 148 brain regions. This information was then classified by a machine-learning model trained to differentiate between brain imaging features in ADHD and typically developing groups [support vector machine-recursive feature elimination (SVM-RFE)]. The authors also analyzed differences in brain organization between the ADHD and typically developing groups using a specialized statistical method for assessing how gray matter structure co-varies across brain regions and individuals (i.e. structural covariance analysis). Finally, each participant was tested for the COMT Val158Met polymorphism, a COMT mutation previously implicated in ADHD, and the resulting four groups (ADHD/mutation-carrying, ADHD/normal COMT, typically developing/mutation-carrying, and typically developing/normal COMT) were compared for neuroanatomical and behavioral differences.

What did they find?

First, the authors found that the cortical thickness classifier (a model using cortical thickness to classify differences between groups) had the highest rate of accuracy (79%) compared to the surface area (74%) and volume (66%) classifiers, suggesting that cortical thickness may be a better predictor of neurodevelopmental disorder than cortical volume. They observed increased cortical thickness in the ADHD group across 16 brain areas relative to the typically developing group, with the most robust increase identified in the orbitofrontal cortex, a region involved in executive functions such as decision-making, reinforcement learning, and working memory. In contrast, cortical surface area was decreased in the ADHD group across 11 regions primarily throughout the frontal and temporal lobes. After further dividing of the two diagnostic groups based on COMT genotype, they next found that the mutation in ADHD subjects accounted for reduced cortical surface area, specifically in the left orbital sulcus (OrbS) (in the orbitofrontal cortex) compared to ADHD subjects with no COMT mutation. In typically developing participants, however, the COMT mutation was associated with decreased cortical thickness in the left orbital inferior frontal gyrus (oIFG) (also in the orbitofrontal cortex) relative to the three other groups. Finally, both the OrbS and oIFG measures negatively correlated with working memory score in ADHD children carrying the COMT mutation.


What's the impact?

This is the first study to use brain imaging techniques to examine the effects of a COMT mutation on cortical thickness and surface area in ADHD vs. typically developing youth. Taken together, the results suggest that abnormal development and maturation of the frontal cortex — as in the case of individuals carrying a COMT mutation — may account for some of the differences in working memory observed between children with and without ADHD. More broadly, these findings present an improvement in understanding how gene function and brain structure interact in neurodevelopmental disorders.  


Jung et al. The Effects of COMT Polymorphism on Cortical Thickness and Surface Area Abnormalities in Children with ADHD. Cerebral Cortex (2018). Access the original scientific publication here.

Spatial Navigation is Guided by an Internally Organized Sense of Direction

Post by Shireen Parimoo

What's the science?

Spatial navigation is a complex process that requires both an egocentric and an allocentric awareness of the environment. An egocentric perspective refers to our knowledge of the environment in relation to our body position, whereas an allocentric perspective is our awareness of the relations between different parts of the environment (e.g. the relation between two buildings). As such, having a sense of direction is important for successful navigation. Head direction cells are a subpopulation of neurons in the medial entorhinal cortex (MEC) that are active when an animal’s head is facing a particular direction. However, it is unclear whether their direction-specific activity is consistently modulated by allocentric aspects of the environment (e.g. the location of two buildings) or if it is internally referenced (e.g. based on our previous position in space) during navigation. This week in Neuron, Park and colleagues investigated how head direction cells represent directions while rats navigated a rotating and stationary arena.

How did they do it?

Thirteen rats were implanted with recording electrodes and habituated to a large circular arena that was either stationary or rotated every 30 minutes. There were spatial cues in the arena (local, olfactory (smell) cues) and in the room that the arena was housed in (distal, visual cues). This was followed by a place avoidance training phase during which rats received a shock if they entered a “shock zone” in the arena. This shock zone remained stable when the arena was stationary and could be identified in relation to distal room cues (e.g. the part of the arena facing the clock on the wall). When the arena rotated, the shock zone dissociated into two shock zones and created two spatial frames – a “stationary” shock zone (room frame) and a “rotating” shock zone (arena frame). In the room frame, the stationary shock zone was the same area that resulted in a shock when the arena was stable (i.e. the part of the rotating arena that faced the clock in the room). In relation to the arena floor, the shock zone would rotate along with the arena. If the rat remained still in one part of the arena, it would only receive a shock when the rotation caused it to face the clock. The rotating shock zone remained fixed to the arena surface but changed in relation to distal cues, so during rotation there was both a rotating and a stationary shock zone.


Neural activity was recorded from MEC neurons during the stationary (two sessions) and rotating phases (one session) of the place avoidance task. The authors identified 115 head direction cells and examined their activity during the place avoidance paradigm to determine spike activity, directional tuning (a cell’s preferred direction), directional strength (the strength of directional tuning), and directional stability (the consistency of the cell’s preferred direction) across the stationary and rotating sessions. They also estimated whether the cells represented internally referenced head direction by examining the firing rate distribution of head direction cells in relation to the activity of one directionally tuned cell rather than the rat’s actual head direction, and they measured the strength of this internally referenced directional tuning in the rotating condition.

What did they find?

The rats successfully avoided the shock zone in both the stationary and the rotating sessions, indicating that their spatial navigation ability was not affected by the rotating arena. However, the directional tuning of their head-direction cells decreased in the rotating condition, and there were more decoding errors when the authors attempted to decode direction from cell activity in the rotating condition versus in the stationary conditions. The directional stability of the head-direction cells also decreased (i.e. became more unstable) across the two stationary sessions, and across the stationary and rotating sessions, than it did within the first stationary session. Specifically, the preferred direction of head-direction cells changed the most in the rotating arena frame condition.

In the rotating condition, directional tuning of cell ensembles was organized based on the spatial frame of the distal shock zone. This means that when rats were close to the room frame shock zone, cell activity and the preferred direction was based on the arena frame shock zone, and the opposite pattern was observed when the rats were close to the arena frame shock zone. These findings suggest that head-direction cells represent directions based on an internally-organized framework, rather than environmental landmarks. At short intervals, the strength of this internally referenced directional tuning did not differ in the room and arena spatial frames. However, after 10 seconds, directional strength decreased in the arena frame, indicating that head-direction cells represent room frame directions more consistently for longer periods of time. Finally, decoding of directions from cell ensemble activity was better in the room frame when the directions were aligned with specific room frame landmarks that did not rotate, suggesting that although these head-direction cells are largely internally referenced, they can also register to directions set by environmental cues.

What's the impact?

This study is the first to show that the head-direction cells in the rodent MEC that represent the internally referenced direction sense also register to environmental landmarks variably and transiently during navigation, instead of operating like a GPS with a stably registered compass. These findings highlight that navigating external spaces is fundamentally egocentric, but the internal direction sense also routinely registers to the allocentric landmarks of the environment for successful navigation. This has important implications for understanding the spatial representations of the environment, which seem to arise from internally-organized, dynamical neural activity that is projected onto the environment, rather than the brain passively reflecting stimuli in the environment.


Park et al. How the internally organized direction sense is used to navigate. Neuron (2018). Access the original scientific publication here.