Temporal pattern separation as a measure of successful cognitive aging

Ooh Ooh Ooh, They Finally Did It!

This post is about a research article I knew existed because I know the authors and I reviewed it. I have just been waiting for it to come out so I could finally celebrate the work. This research finally looked explicitly at temporal processing in humans using a paradigm that is similar to those used in rodents as well as controls for duration as well as sequential order. The manuscript I am describing is “Temporal discrimination deficits as a function of lag interference in older adults” in Hippocampus and is by Jared Roberts and colleagues in Mike Yassa’s lab now at UC Irvine.

An Innovative Approach!

What excites me about this research is that previous work looking at temporal processing in humans has ignored the rodent research and computational models that rely on rodent research for support. What the authors did in their experiments was to specifically design an experiment to test two things: 1) They are testing the potential role for an inability to gauge duration of time between stimuli in aging. And 2) they are looking at specifically whether or not there is a specific deficit for temporal ordering (or sequential ordering to be more precise) in aging.

They found that there was a relationship between the number of intervening items (“temporal lag”) and performance with age. Older adults showed deficits for task performance at medium and high temporal lags, but not low temporal lags. When they looked at potential confounds of primacy and recency, they found that this measure actually separated the older adults into the aged-impaired and ages-unimpaired groups often reported in studies such as these (in fact, they found that this measure was more reliable than the typical RAVLT used to separate groups into impaired and unimpaired groups in aging studies!). The unimpaired group was 2 standard deviations below the young group whereas the unimpaired group was on par with the young group performance levels.

When they looked at their data again, separating the older adults into the aged-impaired and aged-unimpaired groups, they found that the impaired group showed a global deficit at all temporal lags, whereas the aged-unimpired group only showed a deficit at the middle lag. They interpreted these data as suggesting the aged-impaired group had a more global memory deficit in the temporal domain but also extending beyond (perhaps to prefrontal cortex dysfunction, etc). The aged-unimpaired group, however, appeared to show a temporal pattern separation deficit as has been reported in a number of rodent studies.

As part of their validation, they set out to use the standard RAVLT separation of aged adults into groups, rather than try to explain their exults, I will quote them:

One pertinent question is whether or not standard neuropsychological testing that is commonly used to assess medial temporal lobe and prefrontal cortical function is sensitive to the age-related deficits reported herein. We examined scores on the RAVLT delayed recall (Rey 1941) a word list learning test that is sensitive to MTL dysfunction (Lezak et al. 2004). Although prior studies in humans have used scores on the RAVLT delayed recall as a criterion for dividing aged subjects into AI and AU groups, these studies examined memory for objects independent of temporal information. While the RAVLT assesses word list learning, it does not assess the temporal sequence in which those words were presented. Thus, we reasoned that it may not be sensitive to impairments affecting temporal memory specifically, and as such may not be the best criterion for discerning gross temporal memory impairment from more subtle impairment. We found that although there is an age-related decline in RAVLT delayed recall, consistent with prior work (Stark et al. 2013), there was no difference between the AI and AU groups. This suggests that our temporal memory task may be sensitive to more subtle impairments in memory function than standard neuropsychological testing and may offer potential avenues for diagnosis and prediction of cognitive decline.

In other words, they found their task to be more reliable and sensitive than neuropsychological tests. Which to me is not at all surprising.


Roberts and colleagues developed a task to test temporal pattern separation (as well as primacy and recency) in a cohort of young and aged adults. They were not only able to identify a clear pattern separation deficit in the temporal domain (which makes me very happy as I have been harping on their group to do so for quite a long time), but they also were able to identify successful and unsuccessful cognitive aging as well.

To me, the more important part of their work was that they developed a behavioral task that appears to be a rather sensitive marker for episodic memory deficits that emerge with age (at least the “when” component of episodic memory), and appears also to have a potential diagnostic value for early detection of age-related pathology. The authors obviously need to run a much larger cohort to see what individual differences exist in this task similar to their work with the behavioral pattern separation tasks (mnemonic discrimination tasks as now renamed), but it appears they are off to a great start.


A new way to study memory in Down Syndrome

Why Haven’t They Done That Yet?

This is a strange post because I am talking about someone having done the work I will write about, but I am asking someone to do the work. This is my way of trying to drum up replications and increased n’s. I think as scientists we need to replicate every finding we see in populations with developmental disorders, and then extend the research into different disorders to identify commonalities and differences among each disorder’s cognitive deficits.

Also, I feel it is necessary that we not only identify cognitive deficits in patient populations, but we also try to unpack the nature of these deficits. In other words, we try to understand how their brains work, not just ask if they perform worse than a control group.

This post is about some work from a former collaborator of mine as well as some work my old advisor did in an associated mouse model. Briefly, they showed that children and adolescents with Down Syndrome actually process the visual world differently than the rest of us in a very interesting way that has implications for intervention studies. I will also be asking why has this not been done before and is anyone going to follow up on this and test if the same phenomenon exists in other disorders.

I will talk about portions of the paper entitled Dentate gyrus mediates cognitive function in the Ts65Dn/DnJ mouse model of down syndrome that appeared in Hippocampus by Gen Smith et al. and the paper entitled Remembering things without context: Development matters that appeared in Child Development by Jamie Edgin et al.

What Should Be Studied and Why?

Smith and colleagues did a rather ingenious thing in their experimentations. They looked rather closely at the data they were collecting in the Ts65Dn mouse model of Down Syndrome. They ran a series of experiments and found out that they were seeing effects that made absolutely no sense given what we know about spatial memory and the brain. They were getting what appeared to be object recognition deficits that had to do with impaired dentate gyrus function. For those not in the field, this does not really happen. At least not that we know. Not in a normal brain at least.

Further experiments were required to figure this out and determine if the Down Syndrome brain is fundamentally different in this way.

In parallel, Edgin and colleagues were grappling a similar question in their research with individuals with Down Syndrome. They were seeing eerily similar effects that seemed to make no sense… And this mystery needed to be solved

A Valid Approach and Likely Interpretation

Independently and in parallel, both groups came to a very similar solution. Smith and colleagues proposed that there might be an effect of context influencing the object recognition. In other words, if you run a task wherein you ask a mouse to identify an object it matters if the box is clear or opaque. Let me explain. If you do a simple object recognition task in an opaque box, the Ts65Dn mice can do it, as can wildtype mice. However, if you use a clear box that lets the mouse see everything in the environment, they get confused and no longer can perform the task, but wildtype mice can. This suggests one of two things, 1) that the Ts65Dn mice are not able to focus on the object, but get distracted by the environment. or 2) The Ts65Dn mice actually fuse the object with the environment when in a clear box and it is just too complicated for them to process given they have a messed up dentate gyrus.

Edgin and colleagues have a similar solution as well as a similar interpretation. Edgin and colleagues tested kids with Down Syndrome across ages to see if there was a difference in the development of object in scene learning. In one condition they showed a scene and tested if the kids could remember a scene. In another condition there was an object in the scene and they tested if the kids could remember the object in the scene. the last condition showed an object in a scene, but tested if the kids could remember the object when the test did not include the scene.

What they found was that for the kids with Down Syndrome, there was a deficit for remembering scenes as well as a deficit for remembering an object if it was presented in a scene, but no deficit for just remembering an object. They interpreted this as evidence that kids with Down Syndrome perhaps are processing scenes and objects differently than typically developing kids.

So what do you propose?

What I propose needs to be done, and frankly it amazes me how rarely this line of contact is followed, is to evaluate exactly how developmental disorders change the way the brain processes stimuli. In other words, these two papers were great because they came up with a mechanism demonstrating that the Down Syndrome brain (and Ts65Dn brain) actually works differently than the typical brain. They did not stop at, “Hey look, there is an impairment, lets give them drug X”, they tried to determine the precise nature of the impairments. Were they impaired, as in cannot do it? Or were they impaired as in they do it in a different way that fails our particular test. In the case of object recognition, at least at short delays, it is actually clear that there is no real deficit related to Down Syndrome, the apparent deficit is due to task confounds that were not controlled in earlier experiments.

I hope this line of research continues. Long term, we are not going to be able to medicate every child and adult with every developmental disorder. We are going to have to get creative to design appropriate solutions and teach coping skills to individuals in these populations. Unfortunately, in the present translational research context, we are looking for a targetable outcome that can be used to test drugs. But…we are not taking the effort to actually quantify the differences. Once we understand the differences it becomes possible to design training regimens to help integrate individuals with developmental disorders into the job market, design video game or computerized interventions to help them overcome cognitive weaknesses, and perhaps leverage their cognitive strengths to compensate for their weaknesses.

However, we have to ask the questions and run the experiments before we can do all this. And not very many of us are. Nor does it appear we are going to start anytime soon.