We humans have an impressive ability to plan for the future — lapsed New Year’s resolutions and overambitious project planners notwithstanding. We can make decisions in the present (like bringing an umbrella) that make our lives easier in the future (we stay dry). Among the animal kingdom, this capacity is distinctly rare. Even animals that can use tools, like some primates and birds, seem to do so only in order to solve problems that they face in the present . Mostly problems like “ooh look, a snack! Lemme at it!” *uses stick to retrieve yummy ants*.
How do humans project themselves into the future like that? Paradoxically, it may be our ability to remember the past that enables us to predict what’s to come. After all, to imagine the future with any degree of accuracy requires a well-tuned mental model of the world, one that can be calibrated only through experience. That experience and the updates in our knowledge and beliefs that it triggers must be maintained somewhere in the brain, lest they be lost. Memory is simply the preservation in the brain of learned information, and planning for tomorrow makes use of what’s preserved today. Indeed, “it is accurate prediction of the future, more so than accurate memory of the past per se, that conveys adaptive advantages [of memory]” .
Strangely, this idea comes from research in perception, the process by which the brain translates incoming information from our eyes, ears, and other sensory organs into the mental experience of seeing, hearing, feeling, and so on . To understand the connection, we need to appreciate that perception is hard. So hard that the brain needs to constantly predict what sorts of things we are likely to perceive in the world. Here’s an example: you may have seen this optical illusion before. If you haven’t, then you probably can’t tell what’s so illusory about it. Take a minute and see if you can figure it out.
Give up? The squares labeled ‘A’ and ‘B’ are in fact exactly the same shade of grey. But because of the shadow cast by the cylinder, we find it impossible to see them as such, even after learning about the illusion.
What our brains are doing in this scenario encapsulates Carl Sagan’s famous quip, “extraordinary claims require extraordinary evidence” in order to convince you that your preexisting beliefs are incorrect. Take the claim that the Earth is not flat, but round, despite all the evidence of flatness provided by our day-to-day experience. That’s pretty extraordinary! Everywhere we look, the surface of our planet seems flat. If someone clever points out that ships hulls disappear over the horizon before their sails, that’s some small evidence in favor of roundness, but perhaps not enough to be convincing. But that observation plus round shadows cast on the moon by the earth during lunar eclipses, plus a rotated perspective on the moon and stars from the southern versus northern hemisphere, plus mountains of historical and modern data … that does the trick.
In the case of the checker shadow illusion, light hitting our retina provides evidence that the two squares are the same color (because they are in truth the same color, and light can’t lie). However, through evolution (and experience), our brains have learned that when objects cast shadows, they make whatever the shadow hits seem darker, but do not change its true color. Combined with the surrounding checkerboard context, we imagine square ‘B’ to be much lighter than square ‘A’, because our visual system tries to predict their true colors, and if this setup were in the real, 3D world and not on a screen, ‘A’ would be dark and ‘B’ light. The evidence provided by our retina that the squares are the same shade is not extraordinary enough to our visual system to convince it that it’s prior beliefs about light, shadows, and checkerboards are incorrect, and so we inevitably see the illusion. However, a visual demonstration like the one below can overcome the extraordinary burden of proof required to reject the conclusions of our visual system and believe — though still not perceive! — the squares to be the same color.
Now, what does all this have to do with memory?
In perception, our brains’ predictions about what’s likely to be out there in the world are combined with new information from our senses to form updated beliefs about what we’re currently seeing (or hearing, etc.). Perhaps we can think of memory as solving a similar inference problem. When we find ourselves in a situation similar to one encountered in the past (say, riding the subway), our brains may automatically begin using that similar experience to form expectations about what is likely to occur in the present (like elbows, claustrophobia, unintelligible announcements, and the occasional performer). Doing so would help us mentally prepare for what our experience predicts life will throw at us.
Much of the value of prediction may come from getting things wrong. As with the checker shadow illusion, our brains’ expectations are not always met. Sometimes we predict something that fails to happen (today the subway is conspicuously empty), while other times we fail to predict something that does happen (say, a minor train derailment). In either case, that failure of foresight is a good signal that something about our understanding of the world was a bit off, and that we should update our predictions for next time, in an attempt to be less wrong in the future.
Recent evidence is beginning to bear out this idea. Some work on human memory for faces and scenes suggests that when we strongly expect to encounter a particular face in a particular scene, but do not, the memory linking that item to the scene is greatly weakened. Conversely, the unexpected appearance of a particular face in a scene results in a large update to the strength of the memory for that face-scene pair . Other research has detected similarities in the brain activation patterns of people asked to recall experiences in their past and to imagine their future, suggesting that some of the same neural systems are involved in both processes . Finally, patients with memory disorders are likely to report episodes from their past when asked to imagine their future, and often have trouble planning for the future . Taken together, this work hints that memory may really be more about the future than the past. By constantly updating memories to generate more accurate predictions for the future, we may learn to better understand our environment and use it to our advantage.
Although the idea of memory-as-prediction-engine is still relatively new, it might be able to unite two impressive human capacities: planning for the future and remembering the past. What new insights might this perspective offer in understanding human memory phenomena? Whatever the answer, we’ll arrive at it using science. That is, we’ll make predictions, then update our expectations and beliefs based on how wrong those predictions were, which generates new predictions… sound familiar?
 De Brigard, F. (2012). Predictive memory and the surprising gap. Frontiers in Psychology, 3, 703–719. https://doi.org/10.3389/fpsyg.2012.00420
 Greve, A., Cooper, E., Kaula, A., Anderson, M. C., & Henson, R. (2017). Does prediction error drive one-shot declarative learning? Journal of Memory and Language, 94, 149–165. https://doi.org/10.1016/j.jml.2016.11.001
 Kim, G., Lewis-Peacock, J. A., Norman, K. A., & Turk-Browne, N. B. (2014). Pruning of memories by context-based prediction error. Proceedings of the National Academy of Sciences, 111(24), 8997–9002. https://doi.org/10.1073/pnas.1319438111
 Mullally, S. L., & Maguire, E. A. (2014). Memory, imagination, and predicting the future: A common brain mechanism? Neuroscientist. https://doi.org/10.1177/1073858413495091
 Schacter, D. L., Addis, D. R., & Buckner, R. L. (2007). Remembering the past to imagine the future: The prospective brain. Nature Reviews Neuroscience. https://doi.org/10.1038/nrn2213
 Suddendorf, T. (2006). Foresight and Evolution of the Human Mind. Science, 312(5776), 1006–1007. https://doi.org/10.1126/science.1129217
 von Helmholtz, H., 1866. Concerning the perceptions in general, 3rd edn. Treatise on Physiological Optics, Vol. III (translated by J. P. C. Southall 1925 Opt. Soc. Am. Sec- tion
Title image: https://www2.gmu.edu/news/317086
Checker shadow: https://commons.wikimedia.org/w/index.php?curid=4443183
Checker shadow proof: https://commons.wikimedia.org/w/index.php?curid=45737683