Can We Really Test Math Skills in Mice?

Why haven’t They Done That Yet?

I have long been interested in making mouse and rat models more relevant to the study of human disease. My approach has always been to try and increase the level of homology between the behavioral tasks used in humans and the models. I consider this essential since the study of animal models of human disease should directly inform the study of patient populations.

The approach of rigorously translating the cognitive phenotype in patient populations into the mosue model is in its relative infancy and has been coined behavioral endophenotyping (Link, Link). This post is going to discuss the possibility of studying processes associated with math in mice. This is not to say I believe mice will be doing long division anytime soon, but we can test the different processes known to underlie math skills in humans-in the mouse model (magnitude comparison and enumeration).

What Should be Studied and Why?

The reason it is important to study numerical competency in mouse models of disease is the fact that math deficits are relatively common among neurodevelopmental diseases. For example, fragile X syndrome, William syndrome, 22q11.2 (Velo-Cardio-Facial Syndrome or DiGeorge syndrome) all show profound deficits in math and brain function related to learning numbers and magnitudes.

Recently, a series of studies were performed in a population with mutations that expand across generations toward full fragile X syndrome. These studies involved processing of magnitudes using size or numbers as stimuli. These studies found these magnitude processing deficits were present in seemingly cognitively unaffected participants (Link, Link disclaimer: My wife is an author on these two studies).

At present, no good mouse or rat homologues to these types of magnitude comparison tasks exist. The lack of homologous tasks means that we may not be utilizing our models as effectively as we might. I propose that there is a simple way to test these processes in rodents, but it has not yet been attempted to the best of my knowledge.

A Proposed Approach

So I have only been working on this problem in cerebro so there are no data that I know of to support any of these ideas, but I am going to lay out a potential method for testing numerical and magnitude estimation abilities in mice and rats. Eventually I plan to get around to actually seeing if this works, but I feel the need to get the ideas out there now so perhaps someone smarter than me can get a head start.

The human studies I mentioned above used computer-based testing for their tasks. The stimuli looked something like this:

Human Figure 1

Analog magnitude comparison. On the left is a difficult condition and the right is an easy condition. The participants are asked which of these bars is longer. Similar tasks have been performed looking at numerical magnitude comparison (i.e., different numbers of stimuli rather than size of a bar comprise the “magnitude”).

In both of these tasks the participants had to choose either the longer of the two stimuli or the one with the greatest number of stimuli. Importantly, as the size or quantities of the stimuli became more similar, performance degraded as one would expect.

What amazes me about the fact that this has not yet been done, is that there has been a huge increase in recent years in the number of researchers using operant chambers with touch screens attached. Since I mentioned an inexpensive way to make a home-cage bases system in an earlier post, I will use this apparatus to describe my proposal.

I propose that one can program an app on an iPod touch or an iPad mini to show stimuli like this example for a numerical estimation task:

Mouse Figure

For this example, there are two sets of stimuli on the screen. The mouse is simply trained to touch the stimulus block with a higher or lower number of items with a large separation (i.e., 4 vs 7 or 4 vs 1). Once learned, testing would commence with trials that range from very easy (separation of 4, to very hard, separation of 1). Identifying there different mice fail to perform nominally would potentially index the effects of the mutation on the mouse’s ability to process numerical magnitude. This can be made independent on total spatial extent of stimuli by making the total amount of space covered by the stimuli constant across trials.

The same could be done for stimuli of different lengths or widths, with training on the stimuli with the largest size difference and training with all conditions from very easy to every hard.

Potential Interpretation


p>The results one may obtain from experiments such as these would likely serve to better model human populations, particularly those with neurodevelopmental disorders. Of particular importance is if studies like this can uncover some sort of behavioral risk profile that may predict later disease onset or progression. Such prodromal findings may then be used as a sort of behavioral biomarker that can be targeted in preclinical drug or cognitive training studies prior to off-label or clinical drug trials in human populations.


I would love to hear your thoughts on this!

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s