I don’t remember: Why your memory isn’t nearly as good as you think it is…

Chances are, you probably think that your memory is either pretty good, or really bad. You also probably think that (despite your own experience), human memory is like what Hollywood would have you believe: everything you experience is encoded with perfect accuracy, and can be recalled perfectly under hypnosis, torture, or in a crisis. We’re conditioned to believe that human memory is like that of a computer, and we couldn’t be more wrong.


Psychologists have been researching memory since the 1950s, and we’ve known for some time that our memories are shockingly poor. Elizabeth Loftus, a pioneer in the study of human memory, demonstrated as early as the 1970s that human memory is highly fallible, especially in the recall of important facts (such as during eyewitness testimony). She clearly demonstrated that people’s memories of events could be altered substantially through minor interventions, such as the use of descriptive adjectives in prompting a person’s recall (something done by lawyers since the invention of the legal system). For example, in a classic study, volunteers watched footage of a minor collision between two cars. In reality, no glass was broken but, a week later, when participants were asked to recall whether or not they saw broken glass, they were significantly more likely to remember seeing it when asked something like “was there broken glass when the cars smashed into one another”, versus “was there broken glass when the cars bumped one another”.

Ongoing work has shown that eyewitness testimony, police interrogations, torture, and other attempts to gain reliable information based on people’s memory of events, are more often than not, completely useless. Worse, it’s common for people (without realising they’re doing it) to make up new memories to suit the circumstances. Sadly, the legal system still places an undue emphasis on human recall (partly because it’s scientifically illiterate, and partly because it doesn’t have anything better to rely on).

In fact, human memory is a lot more like a game of Chinese Whispers than that of a computer. First, any initial encoding of memory is highly dependent on our level of attentional focus, our mood, our prior experiences, and our priming (subconscious expectations based on events around us). Second, whether or not a memory is stored longer-term is largely dependent on ongoing practice (as anyone who’s studied for an exam will attest to). So the stuff that happens around us is seldom encoded unless we have a reason to do so (in that it stood out in some way, or that we found it personally meaningful or important). Next, each time we recall an event, the accuracy of that recall is also affected by our mood, environment, and level of attention; depending on the people around us, our feelings at the time, or the reasons we have for recalling a memory, the accuracy of the memory will change (usually for the worse). We then reencode that memory based on our impressions at that point, resulting in an edited version that overrides the original. Thus, by the time we’ve recalled something several times, it no longer resembles the original very much, but (and this is the worrying part) we believe that it does. In fact, most people place an annoyingly stubborn faith in their own memories versus those of others. Again, if you’ve ever had an argument with your spouse over who last took out the rubbish, you will have experienced this phenomenon (and you’re probably still convinced that your recall is more accurate than your partner’s). This last point is worrying, because it allows us to place astoundingly large amounts of faith in our own bullshit. It’s also why people go to great lengths to convince themselves and others that they were right, even when there’s convincing evidence that they were wrong (Donald Trump and Cardinal George Pell are great examples of this – in Trump’s case it’s also an example of how narcissism and sociopathy have made him into a liar who doesn’t believe he’s lying – read here). In fact, it’s quite possible (and common) for us to completely invent false memories that we genuinely believe happened – and the scary bit is that we can’t tell the difference.

In other words, not only are our memories poor, unreliable, and often fictitious, we believe that our recall is both accurate and better than those of others – despite being completely wrong a lot of the time. This makes us pretty dangerous when we’re put in situations where that recall can affect others.

Before we go any further, I need to point out that humans have several types of memory, some of which are more trustworthy than others. So far, we’ve been talking about explicit memory, a conscious system that utilises a part of the brain called the hippocampus. This system is the one you and I use throughout the day in order to recall facts, and to structure our experience of consciousness. Using this system is familiar: when you search for and recall a memory, it’s pretty obvious that you’re experiencing a memory – it feels different to immediate experience, and it usually has a sort of ‘timestamp’, in that you can remember when and where it was encoded. Although this system unreliable, it’s remains extremely useful because we can program a lot of effective information into it. Sadly, a lot of what you put into it will decay without regular reencoding (i.e., recall and practice) – which is why it’s unlikely that you can remember much of your high-school maths unless you use those skills in your everyday life (but the bias around the quality of our own memories will make you swear blind that your memory of the thing you did with a friend 15 years ago is more accurate than hers).

In fact – our memory systems work something like this: at the most fleeting level we have sensory memory that lasts less than a second (this is our ‘buffer memory’ for sights and sounds). Above this is our short-term working memory, which lasts for about a minute (this is why you can’t remember why you walked into the living room, or forget someone’s name less than a minute after they tell you). If we rehearse the information held in short-term memory, it is more likely to be transferred to long-term memory for later use but, as I talked about above, this process is fraught with errors. To make things more complex, our long-term memories are two-fold. First, we have an explicit, conscious component (see above) that is divided into declarative memory (facts and events) which, in turn, is divided into episodic memory (events and experiences) and semantic memory (facts and concepts) – as I’ve mentioned, explicit memory is highly prone to errors and distortions in both encoding and recall. We also have an implicit long-term memory system that is handled by different parts of the brain. Implicit memories are unconscious (that is, they happen at a level that we’re not aware of, and have little to no access to), and include procedural memories (like skills and tasks), trauma-based memories (automatic encoding of traumatic events), and schema memories (early experiences based on our upbringing).

Implicit memories are another kettle of fish altogether, in that they aren’t conscious, and can manipulate our behaviour without our being aware of them. Procedural memories are less like memories, and more like instruction sets (actually more like computer memories). Handled largely by the cerebellum (a part of our hindbrains), procedural memories take ages to program (through large amounts of repetition) but, once encoded, are largely automatic and resistant to decay. Driving a car or riding a bicycle are great examples of procedural memories: they take a while to learn, but once learnt are likely to remain indefinitely and recalled without effort. Other types of implicit memories (managed by the amygdala) are usually encoded in order to keep us away from danger (trauma-based memories), or to help us function as children when we’re deprived of emotional needs. Because a lot of our conscious behaviour is manipulated by these memories (without us knowing anything about it), they’re pretty important – but they really require a whole post to themselves – so my next post (called Demystifying the Unconscious) will focus on how our implicit memories shape our behaviours, and why this can be a problem.

Coming back to our explicit memory systems: we’re stuck with a flawed system attached to a delusion engine. We honestly believe that our memories are accurate and reliable, despite the fact that they are repeatedly demonstrated to be poor, biased, and variable. The fact that we (paradoxically) expect others to remember things that, for the most part, they have no chance in hell of being able to do with any degree of accuracy, compounds our belief in the viability and worth of human recall.

There is, however, a good alternative. In fact, I have two mediators for you. First, we live in an era of almost ubiquitous recording, attached to artificial memory systems that are substantially more trustworthy than our own*. Taking advantage of these systems to augment our own, flawed apparatus, can help. For example, I outsource most of my declarative memory to my phone: it remembers where I need to be and when, what I need to get done, how to contact people, how long a task or event took me, and how to get from place to place and, so long as I’ve backed it up well, it’s a lot more reliable than I am. It also remembers what I said, to whom, and in what context (when I use it as a recording device), with little chance of subjective bias or recall error. Utilising our devices appropriately to adjust for our own limitations, can make life a lot easier. Of course, the term “garbage in, garbage out” comes to mind – making sure that the information we enter in the first place is reliable, is pretty important in this context.

Recognising that our ability to recall with accuracy is a lot worse than we think is a pretty good first step – and, instead of automatically trusting our own memories, learning to step back and check with our digital assistants can serve us quite well. Nevertheless, there’s a lot to be said for learning to pay attention whilst understanding the limits of our own abilities. Actually noticing what’s going on in our lives, rather than wandering around in a blind fog of distraction, can make it more likely that the quality of our stored memories will be a lot higher (again “garbage in, garbage out”). Moreover, instead of constantly distracting ourselves with our devices, our time pressures, our worries, our feelings, our stresses, our preconceptions and biases, and our needs for validation, we could spend some time learning to observe both ourselves and our world with a bit more prudence and circumspection. Instead of assuming, parsing, and hitting ‘like’ buttons, we could take the time to think, learn, and question a tad more. Maybe then we wouldn’t be quite so full of shit.


* Note that I’m fully aware that computer systems are also vulnerable to manipulation, and the consequent danger of placing all our trust in those systems. Ah, irony.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.