How long is short-term memory?

Shorter than you might think.

Learning Scientists blog

Yana Weinstein

When I (or any cognitive psychologist) refer to “short-term memory”, we’re talking about memory that lasts for 15-30 seconds. Not minutes, not a day, not a few weeks. Just 15-30 seconds.

This differs quite drastically from the way people commonly use the term “short-term memory”. Have you heard people referring to how they can never find stuff they’ve left around the house, and following this with “my short-term memory is really bad”? According to cognitive psychology, that would be a completely inappropriate use of that term.

Dory claims to have short-term memory loss to explain why she can’t remember which way the boat went; but if she truly had short-term memory loss, she wouldn’t even be able to have the conversation. In general, when anyone refers to memory loss (formally known as amnesia), they are actually talking about long-term memory.

So, cognitive psychologists divide memory into the first 15-30 seconds, and they call this short-term memory, and alllllll the rest of memory that lasts beyond 30 seconds is long-term memory. Why would we make such a skewed split, and why aren’t there more categories (like medium-term memory)? And what good is this type of memory process, if it falls apart after much less than a minute?

You might wonder, what do we use short-term memory for? Even though short-term is very short, you are, in fact, constantly using it. You use it to remember the beginning of this sentence as you get to the end. You use it to sustain a conversation, which involves listening, formulating what you are going to say, and then saying it. You use short-term memory when you are baking, to remember the quantity of flour you need to weigh out. Your waitress will use her working memory to write down your order as you’re speaking it – but note that if she takes the whole table’s order and then goes to the machine to punch it in, she’s probably transferring your order to long-term memory!

The reason why cognitive psychologists believe that there is something truly special about the 15-30 second range that can be separated from all other memory beyond that timeframe is that patients who present with apparently total memory loss are still able to keep things in memory for 15-30 seconds.

The first patient to demonstrate a profound loss of long-term memory along with perfectly intact short-term memory was called H.M. He was treated for epilepsy when he was in his 20s; since this was the 1950s and they didn’t know any better, the doctors removed part of his brain as an attempt to cure him of his fits. This did result in improvement in terms of epilepsy, but with huge consequences: H.M. also lost the ability to form new long-term memories. For the 40 years that he lived after his surgery, he didn’t form any meaningful new memories about his life.

If asked what he did yesterday, H.M. didn’t know, and if asked when he started suffering from memory loss (yes, he knew that something was wrong), he would say maybe a year, regardless of how many decades had passed. When he saw the researcher who tested him probably at least once a week for those 40 years, he would introduce himself anew every time. This is all to say that despite how severely his long-term memory was affected, his short-term memory remained just as good as mine, or yours. That is, if you read out a phone-number to H.M., he could repeat it back to you just as well as the next person. This also explains why H.M. was actually able to hold relatively normal-seeming conversations, as long as the topic did not extend beyond the present situation.

H.M. died in 2008, and the researcher who had studied him for his entire post-operative life recently came out with a book (1) about him (and then, sadly, also died). In this book – which I highly recommend – she uses a metaphor to distinguish between short and long-term memory, which I think is brilliant and worth repeating. Suzanne Corkin describes memory as if it were a hotel, with short-term memory represented by the lobby, and long-term memory represented by the guest rooms. Here is what she says about H.M.:

“The information could be collected in the hotel lobby of Henry’s brain, but it could not check into the rooms” (p. 53).

To illustrate how sensitive H.M.’s memory is to this distinction between short- (15-30 seconds) and long-term memory, consider the following experiment (2). H.M. was shown two shapes, one after the other, and his job was to indicate whether the shapes were the same, or different. The length of time between presentation of the two shapes varied between 15 and 60 seconds. Here are examples of same and different shape pairs:


Images from Prisko (1963)

Images from Prisko (1963)

Let’s first look at how normal control participants, without memory loss, would perform on this task. Imagine that there are 12 trials in the experiment: 6 where the pairs are the same, and 6 where they are different. If you were randomly guessing, you might expect to get about half of them correct. A typical adult will only get 1 of the 12 trials wrong (getting 11/12 correct) when the first and second shapes are presented 60 seconds apart.

Now, what about H.M.? His data appear on the graph below.

Image adapted from Prisko (1963) Figure 12

Image adapted from Prisko (1963) Figure 12

The dot to the far right, above the 0, demonstrates that when the shapes were presented at the same time (as they are in the example above!), H.M. was pretty good at determining whether they were the same or different. He only made one error in 12 trials. When the shapes were presented 15 seconds apart, he was still pretty good at the task (the mean error is between 1 and 2 trials, because these data are actually aggregated across 5 different versions of the task). But once the delay was increased beyond 30 seconds, we see a sharp rise in the number of errors he makes, and by the time the delay is 60 seconds, it’s as though he is randomly guessing. Recall that the average normal participant only makes 1 error when the shapes are 60 seconds apart; this is not a difficult task. What this shows is that while H.M. could keep the first shape in his memory for about 15-30 seconds, beyond that, it faded away, and so he was no longer able to compare it to the second shape and make an accurate decision about whether they were the same or different.

So, remember: short-term memory is like a “hotel lobby”. Information passes through it, but it doesn’t stay for long, and there’s a limit to the amount of information that can fit in.

By the way: you may not be surprised to learn that even before any data were gathered to support the idea of a distinction between short and long-term memory, back in the 19th century William James (the Father of psychology) proposed the distinction. He referred to the two types of memory as primary memory (things that you are holding in your memory right now) and secondary memory (everything else, stuff that you remember for longer than just in the moment).


(1) Corkin, S. (2013). Permanent present tense: The unforgettable life of the amnesic patient, HM (Vol. 1000). Basic books.

(2) Prisko, L. H. (1963). Short-term Memory in Focal Cerebral Damage. McGill University. Retrieved from

From the Learning Scientists Blog