I have grave concerns about the state of behavioral neuroscience. I’ve always been something of a behavior snob, having studied under the extraordinary tutelage of some excellent psychologists (including Gerard Martin & Ginny Grant at MUN and, of course, Dave Mumby at Concordia) who concentrated terrifically on parsing our rats’ behavior and our interpretations of it. We were extremely cautious and thoughtful about the results of our behavioral tests and ran numerous ‘probe’ tests to clarify the contents of rats’ memory and strategies for solving the problems we presented them with. My honors thesis, supervised by Martin & Grant, contained more days of probe testing than days of training trials! This was perfect for me. I loved behavior, the richer more complex the better. But behavior is messy and frustrating and can be a statistical and interpretational nightmare. And for a behavioral neuroscientist it is rather challenging to attribute that mess to any particular brain structure or function. Hence, it seems, we have, in this field, the depressing compression of all the awesome behavior into tiny awful overly simplistic versions of its former self. Herein lie my grave concerns.
When I started my postdoc at Duke, I remember hearing someone complain about a neurobiologist who had mice that had been genetically altered in some way and they were seeking help from the psych department for them to run “the memory test” with the mice. As if there was some test that we could simply administer in a day and hand back a report of “memory”. Why yes, your mouse is without memory. Okay, we could probably establish that. But the nature of the memory’s contents, the kind of memory being evaluated, and the manner in which the contents of memory were acquired and probed matter enormously. As do the emotional, motivational, physical states of the animal! Probably metabolic states too. Vision! Olfaction! Balance? I think my point should be obvious: deducing all of memory to the outcome of one task is dangerous. To a behavior snob, it is also horrifying. Of course all those states plague us no matter what rich and wonderful behavioral assay we use but most of us trained in the art and science of behavior are at least aware of the dangers, carefully consider their impact, and devise ways to rule out or minimize their influence.
The test most often referred to and generally regarded as “the memory test” is one of spatial learning (and sometimes memory) in a pool of water. This is a truly magnificent test that I love to pieces. It is typically called the Morris Water Maze as if a maze is a test and not just an apparatus and while Morris obviously deserves oodles of credit for his work with a pool of water he really deserves the biggest pat on the back for establishing the methods by which to use said pool to characterize the failure in spatial learning and memory with damage to the hippocampus, the part of the brain thought to be the substrate for cognitive maps. These lovely methods work great in rats (not so hot with mice, and boy do we like to put the mice in the pool). Rats (not so much mice) respond appropriately to the features of the task designed to be aversive which propel them to seek out and learn the location of a hidden escape platform. They attend reasonably well to cues outside of the pool (mice are nuts) and come to rely on them to navigate through the space flexibly to find the platform in a specific location. You can see evidence of this learning by taking the platform out of the pool and evaluating the rat’s search patterns. You can “probe” their memory in other ways by manipulating features of your room or procedures. You can also rule out issues with emotion or motivation by making the platform visible or marking it with visible and prominent cue. If the rats swim reliably to it you can be fairly certain that they are motivated to get out of the water and that they can see reasonably well.
These probes are pretty much the most important part of all of this. Without them, who knows what’s up with your animal? Not I, the humble reader. Also, if you conduct a “probe trial” or transfer test in which the platform is absent from the pool and all your rats spend more time swimming near where the platform was more than anywhere else in the pool, they probably remember where it was and are searching for it. However, if one group is really, really, really persistent and swims near it for longer than the others and the others are still swimming there more than other places: they still all remember where it was. Do the really, really, persistent rats remember better?? Does the student who shouts out the answer louder and sooner know the material better than the student who politely raised their hand and waited to be asked to speak? Maybe. Maybe not. (One day soon I’ll write more about my irritation with statistical analyses of these kind of data.)
In the meantime, this “memory test” exists in a zillion different forms and everyone has their own methods: the literature is overrun with variations and methodological noise. As the reviewer on a paper, I asked an author to tell me the intertrial interval they used and was informed that they did not know and it was not relevant. Then smoke came out of my ears. Some researchers like to put rats in the pool over and over until all their trials are finished, that’s about every 30 seconds or so. Some like to run rats in groups and thus separate their trials by 5 or 10 minutes. That is most certainly relevant. It seems I am ranting.
Let’s just talk about some clever ways to use the pool while I calm down:
In a very popular variant of the “Morris Water Maze”, the platform location is held constant over trials and rats learn its location relative cues outside of the maze, including visual, olfactory, and auditory cues. If all you do is put the rat in the pool over and over again, hopefully from different starting points on each trial, they will learn where to find the platform in the pool. This is called allocentric reference memory. It is allocentric because navigation is dependent on those cues outside of the pool and the problem (of finding the platform) cannot be solved simply by following the same path each time or navigating to a visual cue that marks the location of the platform (in which case it wouldn’t be hidden really). It is reference memory because the information needed to be successful (the platform’s location) does not change from trial to trial. If all you do is administer trials to rats using these procedures you will have established spatial learning and you can test memory by placing the rat into the pool at some point after learning was established (could be a day, a week, or much longer) and seeing how easy it is for them to find the platform (if they do).
Another useful variant entails changing the information on each trial. Now we can assess allocentric spatial working memory. It is allocentric for all the same reasons above. It is working memory because the hidden platform is moved on each trial. This requires a “study” swim to teach the rat where the platform is located and a “test” swim to see how well they retain that information over various retention intervals. In this case the required information (the platform’s location) is usually presented rather fleetingly (often just once), in contrast to the reference memory procedures that use many trials to teach the rat the platform’s lcoation. Thus, you cannot use as long of a retention interval as memory should be expected to decay fairly rapidly. Rats can be trained to keep that info in mind for several hours, or even 24 hours, but anything longer is not very reliable. (One day soon I’ll write about the absolute necessity of a zero, or very short, retention delay.)
I am describing these two variants (and there are others) to highlight the shortsighted and frankly asinine notion that there is “a” memory test. Memory is amazing and complex and intricate and the neural processes and systems that support it and aspects of it and versions of it are as amazing and complex and intricate. As a behaviorist I am not done understanding and thinking about memory and as a neuroscientist I am definitely not done thinking about what happens in the brain to support memory and putting these two together most definitely requires my full attention to them both.
Viva la intersection!