Human memory may be unreliable after just a few seconds, scientists find

The idea that human memory is basically unreliable is a debunker myth that is easily refuted on the basis of recent scientific research, and yet it's likely the primary claim debunkers rely on to undermine confidence in the existence of anomalous phenomena. It's kept alive by low-quality experimental psychology work that overstates the significance of human memory errors and biases, generally by subjecting people to some highly unusual and artificial testing conditions designed to inflate rates of error. Look at the study described in the article linked in the OP: https://journals.plos.org/plosone/articl...ne.0283257. In two different experiments participants were sat in front of displays and shown letters and pseudo-letters in rapid succession and were instructed to remember what would subjectively be small details of what they were shown. Like the many bogus studies of "implicit bias," the modalities of the experiments were such that participants may have been put under significant cognitive load and could have easily become confused, as is made obvious in the paper's methods writeups: "The participant was extensively instructed to not confuse real and pseudo-letters." Even more revealing is that five participants out of 45 had to be excluded from the first experiment because they didn't do well enough in the "training" for the experiment! This is a clear tell that this was effectively designed to trip people up. How informative can something this contrived be about memory in real-world conditions?

Probably not much at all, because such work massively contradicts studies of human memory in naturalistic conditions that find that it is essentially or even highly reliable: https://journals.sagepub.com/doi/abs/10....lCode=cdpa.

Even more impressive are the findings of Nicholas Diamond described here: https://www.psychologicalscience.org/obs...-imperfect.

>Even as the “memory wars” surrounding these and other possible memory errors play out in courtrooms and other venues worldwide, it can be difficult to determine the actual prevalence of misremembering under real-world conditions, Diamond told the Observer.  Diamond and colleagues explored that prevalence by analyzing 74 participants’ recollections of two verifiable events at Baycrest hospital in Ontario, Canada. The first group of 34 participants took part in a standardized mask-fitting procedure for hospital staff during the 2003 severe acute respiratory syndrome (SARS) outbreak. The remaining 40 participants experienced a staged tour of a small art exhibition on the hospital’s ground floor.  In both cases, participants were invited days to years later to share their memories of the event with a researcher who encouraged them to volunteer additional details using general statements such as, “Is there anything else you can tell me about this event?” The memories were recorded and scored based on whether each detail constituted a specific episodic memory and whether the researchers could objectively verify that the event had occurred as recalled.

>Despite the significant decline in the number of details recalled by participants who were older or whose interviews took place later, one finding was consistent across groups: The vast majority of verifiable details that participants freely reported in the interviews were accurate. In fact, the lowest accuracy of any group was 93%, on average, for older tour participants, and no participant in any group was found to be less than 70% accurate. 

>“Consistent with centuries of past work, participants in our study forgot a great deal of information, and they recalled fewer and fewer details with greater retention intervals (e.g., 2 years vs. 2 days) and in older age,” Diamond explained. The accuracy of information that participants did recall was very high, however, suggesting that when people aren’t pressed for details in a leading manner, they’re likely to share only information they are confident they can remember.

I regard these findings as utterly devastating to unthinking debunker bilge about how we can't trust human memory and testimony.

Thanks for this post, @RViewer88. I thought that the debunker claim that memory was unreliable was true, but I agree that these studies shows the opposite to be the case. However, I wonder what you think about these two studies. If I have understood things correctly, they do not use artificial testing conditions but instead uses the real-world condition of major news stories. These two studies still claims that memory is unreliable. I find it difficult to understand why those studies finds memory unreliable while the studies you wrote about finds memory reliable. What is your opinion about this?
Ulric Neisser and Nicole Harsch. "The ecological study of memory". Philosophical Transactions: Biological Sciences. 52 (1362): 1697–1701. https://www.ncbi.nlm.nih.gov/pmc/article...415921.pdf
Jennifer Talarico and David Rubin. Confidence, Not Consistency, Characterizes Flashbulb Memories in Psychological Science Vol. 14, Issue 5, 2003. https://www.academia.edu/314498/Confiden...b_Memories
I found out about these studies in this blog post by Prescott, which discusses the studies: https://michaelprescott.typepad.com/mich...orget.html

As RViewer88 pointed out, such studies greatly exaggerate the unreliability of memory and in themselves exhibit great unreliability. The replication crisis in psychology research is an indication of big problems in this research area.

But anyway I think that this is completely irrelevant to the validity of much of the empirical evidence for paranormal phenomena. For instance, there are the birthmark/birth defect cases where the present existence of certain birthmarks or birth defects in the now mostly grown-up child is a physical fact uncovered by the investigators, as is the existence of the newspaper accounts and medical records that indicate where and how the previous personality in the past life met his or her sudden traumatic death, closely correlating with the present observed birthmarks or birth defects. Memory issues only partially enter into this kind of investigation.

Then there are the many powerfully evidential mediumistic communications investigated in the past, such as with Leonora Piper, where the in-trance utterances of the medium were recorded in writing at the time and later compared to known facts concerning the apparent deceased communicator. And other related experiments such as the "book tests".

Other empirical evidence to which these concerns are irrelevant are the many replicated demonstrations of psychokinesis and/or precognition documented by parapsychologists such as Dean Radin.

Thanks for this post, @RViewer88. I thought that the debunker claim that memory was unreliable was true, but I agree that these studies shows the opposite to be the case. However, I wonder what you think about these two studies. If I have understood things correctly, they do not use artificial testing conditions but instead uses the real-world condition of major news stories. These two studies still claims that memory is unreliable. I find it difficult to understand why those studies finds memory unreliable while the studies you wrote about finds memory reliable. What is your opinion about this?
Ulric Neisser and Nicole Harsch. "The ecological study of memory". Philosophical Transactions: Biological Sciences. 52 (1362): 1697–1701. https://www.ncbi.nlm.nih.gov/pmc/article...415921.pdf
Jennifer Talarico and David Rubin. Confidence, Not Consistency, Characterizes Flashbulb Memories in Psychological Science Vol. 14, Issue 5, 2003. https://www.academia.edu/314498/Confiden...b_Memories
I found out about these studies in this blog post by Prescott, which discusses the studies: https://michaelprescott.typepad.com/mich...orget.html

Looking at the Prescott page, I see a strong emphasis on memory of news items. Here again, it really looks to me as if these psychologists are choosing paradigms for studying memory likely to give high rates of error, probably deliberately even if they pretend otherwise. News items read about or seen on TV, even if dramatic, are not events that are experienced as happening to the individual. They are things out in the world, happening to other people. This study that I already wrote about tested people's memories about a pretty trivial experience, but one that the people actually lived out. They didn't just read about it or see it on TV. That seems like the obvious reason for the difference in memory accuracy. The Challenger explosion study looks to me to be especially idiotic. It was based on "a questionnaire asking about" study participants' "personal circumstances when they heard the news." Look at how trivial the "circumstances" they were probing into were: "there were questions about where they were when they heard the news, what time of day was, what they were doing at the time, whom they learned it from, and so on." I don't find it the tiniest bit worrying as far as human memory goes that people aren't very good at recalling such insignificant stuff a "year and a half later." Notice that they used a questionnaire. They didn't ask people to simply recall what they could remember like the study I reported that had the very high accuracy rates. If you ask people to answer a pre-determined set of questions, they'll guess or strain to come up with something when they don't know.

It's unfortunate that Prescott is relying on the poor and deeply ideological writing of Bart Ehrman for guidance on this topic. The New Testament that Ehrman has wasted so much of his life incompetently and dishonestly attacking contains a huge amount of astonishingly accurate historical information that testifies to the strength of human memory, not its unreliability. For more on that, you can read The Book of Acts in the Setting of Hellenistic History by Colin Hemer. Here there's information on the Book of John, often ignorantly disparaged as the most obviously unreliable of the four Gospels.

Returning to the news story and "flashbulb memory" stuff, not all studies have found evidence of poor reliability of such memories:

https://www.tandfonline.com/doi/abs/10.1...21.1891253

>Nevertheless, we believe that our study provides good indication of high consistency of FBM after more than a year. Indeed, for perceptual elements, participants were generally able to provide an accurate and consistent answer on the canonical questions.

The "perceptual elements" part is key. They imply that ignoring this, and other factors, can lead to serious exaggeration of memory unreliability:

>Yet, perfect consistency highly depends on the type of
questions and the strictness of the coding system. In our
study, participants could precisely recall the exact place
they were in when they heard about the bombings and
the other person/people present at the time, but they
usually were more imprecise or did not know the exact
time they learnt the news. These results fit with the idea
that FBMs are reconstructed with almost perceptual
clarity (Brown & Kulik, 1977). The exact time is not a perceptual
element. One might recall with perceptual clarity
that it was sometime in the morning as they remember
the morning sun shining through their windows, yet
they might not be able to specify whether it was 8.10 or
9.30. It is also harder to be consistent in the way one
reports one’s activity, as already seen in other studies
(Curci & Luminet, 2009; Curci et al., 2001). There are
many ways to express an activity. Someone might say “I
was studying at might desk” at T1 and then explain at T2
that they were “working on their computer on an assignment”.
These responses are not consistent, yet they
could both be accurate. We believe to have reduced this
issue as much as possible by using an interview method
to collect data at T2 and prompt for further explanation
when necessary.

Something else that comes up in the articles you linked and at Prescott's blog has to do with the supposed ease of making people remember things falsely. This also seems exaggerated according to 2021 paper by Alan Baddeley, "Is the study of memory unduly preoccupied with its sins?" which as you'd expect from the title is aware of the growing evidence that psychologists have given us an inaccurate idea of the reliability of human memory by overemphasizing its weaknesses and not appreciating its overall strength. He says this:

>As Schacter’s (2001) book suggests, much research has
been concerned with memory failure, particularly in the
applied field where two topics with potentially important
legal implications have been particularly prominent in
recent years namely the unreliability of eyewitness testimony
and the danger of inducing false memories, particularly
associated with claims of recovered memory in adults
reporting child abuse. These are both important areas
where psychologists have made major contributions
(Davis & Loftus, 2018). However, it has more recently
been suggested that the prevalence of studies, and in particular
those based on laboratory simulation of such effects
may have resulted in an undue emphasis on the fragility of
human memory, its sins rather than its virtues. Recent
studies (Diamond et al., 2020; Diamond & Levine, 2020)
using a large sample study on the recall of natural
events report that, while retention of details declined
rapidly, the accuracy of the information that was still
remembered remained high, (93–95%). They contrast this
with the expectations reported in a survey of 400 academic
memory scientists who predicted around 40% accuracy at
best. In a similar vein, conclusions drawn from studies
aiming to implant false memories designed as an analogue
of potential introduction of false memories of child abuse
have also criticised the recent literature as leading to an
exaggerated view of the magnitude and extent of such
events (Brewin et al., 2020; Brewin & Andrews, 2017).
This conclusion has however been challenged by Scoboria
et al. (2017) who point out the importance of exactly how
memories are classified, they review an extensive range of
studies, applying to them a carefully specified set of criteria
and finding a much higher rate of false memories.
Others have however questioned the criteria used,
suggesting a need to distinguish between genuine memories
and beliefs, implying for example that if the
suggested “memory” has been suggested by an apparently
reliable source, then it must be true, with belief
then leading to construction of a plausible scenario
(Shaw & Porter, 2015; Wade et al., 2018). This is clearly an
important issue, that of how we decide that a retrieved
episode is a memory. This may however reflect a decision
that depends on a range of different features, few of which
are easy to monitor. It is important to note however that
both these and the Diamond studies accept that substantial
forgetting occurs, at issue is the incidence of erroneous
memories of the type that could prove misleading in a
legal context and the extent to which these are typical
of everyday life.

The papers he cites showing false memory claims are exaggerated show that even when extreme measures are used to deceive people into forming false memories only 15% of the attempts result in what participants regard as full memories.

In the paranormal context, Andreas Sommer wrote this in his essay for the BICS contest about memory research:

>Unsurprisingly, however, such public endorsements by James as the leader of the
American psychological profession would only make other psychologists ramp up their
efforts to demarcate their fledgling discipline from politically dangerous associations
with the ‘occult’.165 A comparatively harmless example of such ‘boundary-work’166 was
the appropriation of Hodgson’s study of the fallibility of eyewitness testimony by Joseph
Jastrow, an ‘enlightened’ psychologist on a life-long mission to eradicate paranormal
belief at all cost. Jastrow, America’s first major popularizer of psychology, cited
Hodgson’s findings to imply they demonstrated the wholesale fallibility of testimony for
all psychic phenomena.167

>Jastrow here pioneered a basic strategy adopted by Skeptics and their champions
in academic psychology up to the present day: The application of insights from the
psychology of error without limits on a general, abstract level to explain away any belief in
the paranormal, as an excuse to bypass systematic engagement with the best concrete
evidence and cases.168

>However, psychical researchers were not just the first to systematically formulate
and experimentally demonstrate the fallibility of perception and memory. The primary
sources also show they went out of their way to apply the lessons learnt from these
insights to systematically eliminate or limit errors – to ensure, one could say, the sifted,
published evidence would stand in a court of law. In the case of research on apparitions,
for example, it’s simply not true that psychical researchers were typically satisfied with
cases of ambiguous impressions reported to be perceived in low light, at long range,
fleetingly, or by uncritical people in a state of expectation or similar conditions known to
be associated with pathological and non-pathological hallucinations. On the contrary, the
bulk of published cases has focused on perceptions of vivid apparitions over the course of
several minutes at close range and in bright light, and under such conditions they have
also been reported to be perceived collectively, i.e. by more than one credible witness.169

>Moreover, while Skeptics typically assume that memories of anomalous

experiences are embellished over time to inflate their significance, those who have
actually tested this assumption by re-interviewing recipients found that experiences are
usually remembered consistently, over the course of up to 20 years.170

You can see footnote 170 of his paper which is freely available online to read the papers that support that last paragraph from him.