Skip to content

“Time Travel in the Brain”

Natalie Biderman and Daphna Shohamy wrote this science article for kids. Here’s the abstract:

Do you believe in time travel? Every time we remember something from the past or imagine something that will happen in the future, we engage in mental time travel. Scientists discovered that, whether we mentally travel back into the past or forward into the future, some of the same brain regions are activated. One of those regions is the hippocampus, a brain structure famous for its role in building long-term memories. Damage to the hippocampus causes memory problems, but it also impairs the ability to imagine future experiences. This brain connection between remembering the past and thinking about the future suggests that memory, planning, and decision-making may be deeply related. The ability to form memories allows us to reminisce about the past. But maybe the ability to form memories also evolved to allow us to think about and plan for the future.

And here’s how the article begins:

Let us start with a riddle. Try to figure out how to throw a ping-pong ball so that it will go a short distance, come to a dead stop, and then reverse its direction. You cannot throw the ball such that it will curve back to you (like a frisbee), and the ball is not allowed to touch or be attached to any object. What can you do? The solution to this riddle is described in the footnote below. Yes, it is a simple solution. Do not worry if you did not figure it out immediately, many people do not. The interesting question is why. . . .

And they conclude:

When we think about memory, we usually think about the past. Indeed, for more than a century, memory researchers focused on how people and animals store and recall past experiences and which brain structures support those functions. More recent research suggests a different view of memory. Recent findings show that the hippocampus—a brain region responsible for memory—is active when people imagine future events. Additionally, in patients with amnesia, damage to the hippocampus impairs the ability to imagine the future. Moreover, when rats navigate their environments, neurons in the hippocampus “simulate” future paths that will enable them to get to a desired outcome. Together, these findings suggest that the hippocampus and its connections to other brain regions build upon past experiences to make predictions about future events. . . .

This is interesting, not just for kids. I’m posting it here because Natalie worked on it as a student in our Communicating Data and Statistics class. Previous final projects for that course include ShinyStan.


  1. gec says:

    Thanks, this is a great example of public science communication. It’s very approachable and, despite my best efforts, I can’t find any place where they simplified anything too much that it was incorrect. They got straight to the core of the idea: our experiences of the past, present, and future are similar by virtue of relying on the hippocampus to link things together to form coherent representations of scenes and events, whether current, remembered, or imagined.

    My only complaint is that opening the page for some reason scrolls to the footnote first, which spoils the riddle.

    I was also just telling my mother that she shouldn’t take too seriously anything published in Frontiers and while I stand by that, this at least ups their score in my book.

    • Responder says:

      It’s because the link is to the footnote (the #KC2 at the end of the hyperlink). Removing it takes you to the article head.

    • gec:

      Interesting – any sense why the word simulate was in quotation marks?

      Your wording “coherent representations of scenes and events” seems to suggest it literally is a type of simulation, though maybe only once for each representation…

      • gec says:

        Personally, I think simulation is a reasonable term, though of course figuring out exactly what that means in any particular case is tricky. For example, the “hippocampal replays” in rats are a lot faster than the original neural firing patterns when they were in the maze, so it is not a literal simulation.

        I suspect the quotes are out of an abundance of caution. Terms like “simulation” and “representation” are philosophically loaded (though I’m not a philosopher myself, so I can’t really say why they get so uppity about them).

        Moreover, they only used quotes on “simulation” in reference to studies with rats, so another reason might be that they want to avoid ascribing the same subjective experience to humans as to rats. There’s a long tradition in neuroscience of bending over backwards to avoid ascribing subjective experience to nonhuman animals. For example, many studies of memory in nonhumans will use terms like “episodic-like memory” to distinguish it from human “episodic memory”, even if the inferred mechanisms and behavior are equivalent.

  2. Matt Skaggs says:

    From the article:

    “…certain neurons in the rat hippocampus, which we call “place cells,” play a role in the rats’ memory of specific locations. Scientists recorded the pattern of activity of these place cells as the rats moved around in a maze and again later when the rats were finished exploring the maze. They found that some of the activity patterns measured in the place cells repeated themselves later, when the rats were resting or sleeping.”

    Hey! My brother was first author on that paper!

    “Replay of Neuronal Firing Sequences in Rat Hippocampus During Sleep Following Spatial Experience”

  3. I figured out the riddle within a few seconds–not because I have a lot of experience throwing ping-pong balls into the air, but because it reminded me of similar problems or concepts from elementary physics. The authors write: “Some have argued that people have difficulty solving this riddle because the scenarios we are able to imagine are limited by our past experiences in the world. We are used to seeing ping-pong balls moving around horizontally. So, when we see the ball in our mind’s eye, it is hard to envision it moving vertically toward the sky, because we simply have not witnessed many ping-pong balls flying up and down.” That may be true, but it’s also possible that our mental models–not quite the same thing as empirical experiences–influence how we perceive and solve a problem. (I am thinking of mental models as discussed by Philip N. Johnson-Laird in his book How We Reason.)

    • gec says:

      I agree the riddle may not be the most apt for what they are getting at, but I would say (can’t speak for the authors) that mental models are implemented via a type of “simulation”. This simulation is a mixture of retrieval of prior similar instances as well as more “algorithmic” processing that involves generating (imagining) novel outcomes.

      This basic idea has been prevalent in cognitive science for many years, including Logan’s (1988) instance theory of skill learning, the Exemplar-Based Random Walk (Nosofsky & Palmeri, 1997), and decision field theory (Busemeyer & Townsend, 1993). The main idea behind all these approaches is that you choose what course of action to take by retrieving similar instances from the past; we could call this “simulation”, where the simulation results are based on prior experience.

      A good analogy is using Monte Carlo sampling to approximate a distribution. By sampling from prior experience in proportion to its similarity to the current scenario (the likelihood), you build up an understanding of the range of plausible outcomes (the posterior) which lets you make an informed action. Obviously, this leaves a lot unspecified (how does retrieval work, how is similarity determined, what are the relevant dimensions) but hopefully illustrates how something like a “mental model” might be implemented by sampling from memory.

      • Yes. I suppose there’s an important difference between searching your memory for experiences with ping pong balls in particular, and translating the ping pong ball into something more abstract. Without the abstraction, you might miss some of the possibilities. Abstraction itself is a kind of experience, different from (yet related to) its empirical counterpart.

        • > Abstraction itself is a kind of experience
          For instance if you make an abstraction into diagram you can experiment on that to better understand the abstraction through experience. If you add symbols as well as diagrams some consider that mathematics. Experimenting all the way up.

          • gec says:

            If you folks are interested, I’d recommend checking out
            – Hintzman, 1986, “‘Schema Abstraction’ in a Multiple-Trace Memory Model”
            – Thomas, Dougherty, Sprenger, & Harbison, 2008, “Diagnostic Hypothesis Generation and Human Judgment”
            for how abstractions can be built out of individual experiences and used for various tasks.

            In addition
            – Nelson & Shiffrin, 2013, “The Co-Evolution of Knowledge and Event Memory”
            talk about how those abstractions get stored and used to represent future experiences.

            Finally, Rob Goldstone and his collaborators have done a lot of applied work on how abstractions can be built up from concrete examples
            – Fyfe, McNeil, Son, & Goldstone, 2014, “Concreteness fading in mathematics and science instruction”

  4. John says:

    Wouldn’t “throw the ball at a fan or into a stiff wind” be an equally acceptable answer?

  5. rm bloom says:

    Reminds me of a party trick I used to get remarkable results from.
    Ask some cooperative person to do a long chain of arithmetical computations one after another, i.e.
    “Pick a number; ok? add 12; multiple by 6; subtract 3; blah blah blah….”
    Gone on like this for about 10-15 stages.
    Then you ask them, “Name a vegetable”
    Almost invariably they’d answer “carrots!”
    You then turn around the piece of paper you’ve been holding in front of them the whole time, on which was written the word

  6. Asim Riaz says:

    Personally, I think simulation is a reasonable term, though of course figuring out exactly what that means in any particular case is tricky. For example, the “hippocampal replays” in rats are a lot faster than the original neural firing patterns when they were in the maze, so it is not a literal simulation.

Leave a Reply