Q: My wife and I disagree about memory. She says memories are stored just as they are stored in a computer, meaning that information is placed in some physical form and later retrieved. I say that memory is too slippery for that. If we store memories like a computer, why are memories so unreliable? But if they’re not stored like a computer, where do memories go and how do we retrieve them? Help us settle this once and for all! – James & Kathy, Topeka
A: Dear J&K,
Since you both seem to believe that memories are stored someplace it may turn out that neither of you are happy with my answer. If that’s the case you should feel free to use me as a shared irritant. Camaraderie has many sources.
You’re right, James: memory is often unreliable. In a landmark study on eye-witness reliability, Elizabeth Loftus demonstrated that people can easily be persuaded to recall events that never occurred (Loftus & Palmer, 1974). For example, she showed volunteers a movie scene of an automobile accident and then asked them which car had its windows broken during the crash. In the movie that they saw, neither car suffered broken windows. Nevertheless, perhaps because of the suggestive nature of the question, many people “remembered” broken glass and identified one of the cars.
In addition to having a creative flair, human memory can be spotty, sometimes conveniently so. Who among us hasn’t managed to forget an unpleasant detail or a dreaded appointment?
About that Storage Metaphor…
The storage metaphor (the idea that we store memories in physical form and later retrieve them) is nearly as old as writing instruments. Today we compare our memories to binary information stored on a computer disk; a few centuries ago we were comparing memory to information written on parchment. Whatever the medium, the idea is still the same: we file a memory away in some dormant arrangement of brain cells or chemicals, then retrieve it when it’s needed. In the 1960’s, researchers thought that they had actually pinpointed the location of memory to the chemicals inside cell nuclei but that theory, like many others, didn’t pan out.
The problem with the storage metaphor is that there simply isn’t much evidence to support it, and no one has yet been able to find the location of specific memories. In one experiment, Karl Lashley trained rats to run a maze, then he cut various sections of the rats’ cortex (the outer cells of the brain) hoping to isolate the location of the maze memory (Lashley, 1950). But no matter where he cut, the rats still remembered how to run the maze. Other attempts to isolate memories have been equally exasperating (for us) and unfortunate (for rats).
So what about those brain scans that show different parts of the brain “lighting up” when we are thinking? Don’t they show us the location of memories? Short answer: no. If we are going to stay close to the data, then we can say very little about what brain scan images are showing us. At most, we can point to areas of the brain that become active when remembering, but we certainly can’t say that an image is showing us the location of a memory. While that may seem intuitive, it is only slightly less arbitrary than saying that the image shows us where aliens interface with our brains and tell us what to remember.
When you get right down to it, we simply don’t know very much about the mental functioning that brain imaging technology shows us. All we can say with any certainty is that a brain scan image is a graphical representation of metabolic activity, and nothing more. In other words, brain scans show us behavior.
Remembering Is a Verb, Like Walking, Talking, or Slapping Butts
Thinking, feeling, and remembering are behaviors, just like speaking or throwing a football. You’re saying: “You’re full of it. Even if remembering is a behavior, it still has to draw the memory from someplace.” Not necessarily. Let’s try a thought experiment.
Let’s say that I invent a machine designed to slap butts. When I press a button, a motor activates a few gears and levers and the machine smacks the behind of anyone standing in the target zone. We can say that this machine “knows” how to spank. We can even say that it “remembers” how to spank, otherwise it would not do the same thing the next time we pushed the button. Just like us, it can perform a task on command. The differences between us and this machine are that we have a broader repertoire and we seem to have free will. Beyond that, there is a great deal of similarity: it “remembers” how to do the thing it does, and we “remember” how to do the things we do.
Here’s the brain-teaser: when the machine is resting, meaning that it’s ready to go but the button hasn’t been pressed, where does the memory exist?
As far as I know, there is none, at least in the sense of a “stored” memory. The answer is that even in its resting state, the machine has the potential to smack butts through its arrangement of parts, much as we have the potential to throw a football even when we’re not doing it.
Now you’re saying, “you’re still full of it because human beings aren’t butt-smacking machines.” Speak for yourself, Spanky. I won’t argue that point because I don’t know what kind of machines we are. But people who study memory usually speak in terms of synaptic potentiation rather than storage. Think of a series of neurons as being like people standing in a line. In order to send a message down the line, person #1 gives a high-five to person #2, who then gives a high-five to person #3, and so on. Neurons work in a similar way.
Synaptic potentiation includes the idea that the more you use a given series of neurons in the brain, the more efficient that connection will become and the more likely it is that it will be used in the future. In a similar fashion, the people standing in the high-five line will become more efficient with practice. Not necessarily faster, but more efficient. Synaptic potentiation also includes the idea that we can form new connections between old neurons. (See, for example, Foehring & Lorenzon, 1999, for a down-and-dirty discussions about the biology of potentiation).
So. Strengthening neuronal connections by using them is something like building a butt-smacking machine: the act of putting certain parts in relation to each other increases the chances that a butt will be smacked. If theories involving potentiation are correct, there is no need to have a place to store memories. Why? Try my second though experiment:
Apples to Apples
Look at an object. Let’s say it’s an apple. Visualize what’s going on as you recognize this object: light strikes your retina and the optic nerve to your brain becomes active. The activity from your optic nerves is detected by the visual cortex in the back of your head. The visual cortex interacts with other areas in your noggin until your brain is able to make sense of the situation and remember “apple.” Go ahead and visualize all of this activity any way you want. Imagine the impulses bouncing around inside your skull, various parts of the brain communicating with each other. The specifics don’t really matter for our purposes.
Now, visualize those same mental processes, but imagine them happening when your eyes are closed. Imagine all of the same “apple” impulses bouncing around inside your head. Your eyes are not involved, but everything else is pretty much the same. If you performed the same mental activities without looking at the apple as you did when you were looking at the apple, doesn’t it make sense that some of the same sights, smells, sounds, and sensations of that apple would show up in your mind? If so, you’ve just “remembered” without the need for storage.
In fact, physical storage would add tremendous complication to the act of recalling the apple. In order to store all of that sensory information like a computer, you would have to encode it, label it, file it, retrieve it, and decode it. And you’d have to do that for every apple you’ve ever seen or imagined. Who has the time? If you ask me, potentiation is much simpler.
Of course, I could be wrong, J&K. Feel free to keep looking for that physical storage. I know it’s a hard idea to let go of, and you might not be wrong. I want to be the first to know when you find it.
Foehring, R. C., & Lorenzon, N. M. (1999). Neuromodulation, development, and synaptic placticity. Canadian Journal of Experimental Psychology, 53(1), 45-61.
Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behavior, 13(5), 585-589.