Verified by 
The Headcase is in Washington this week for some World War II research. In several months of interviewing veterans and their spouses, I've learned that sometimes anything can trigger a crisp memory from seventy years back, and sometimes everything cannot.
While talking to a veteran two weeks ago, for instance, I had only to mention the surname of a detested 1st Sergeant for the man to erupt, verbatim, into that sergeant's standard greeting: "I could whip any sonuvabitch in this outfit, and if you don't believe me I'll see you after retreat." * Yet on another recent occasion I spent hour and hour prompting a veteran's widow with precise dates and locations, only to elicit little more than a kindly old smile.
Any journalist knows that memory is a fickle mistress. One trusted way to make an honest woman out of her is with photographs. Something about flipping through a photo album returns us to that unbroken parade of the past. At the very least, a picture lends an element of certainty to highly uncertain proceedings: either you recognize what's going on in the image, or you don't.
At least that's what I've always believed. Turns out that's not entirely the case. A new study raises some important caveats to the accuracy of photo-elicited memories. While past research has found evidence that people will falsely recall words similar to others learned on a list, the current work extends our fallibility to concrete images.
In a recent issue of the journal Memory (pdf here), psychologists Yana Weinstein and David R. Shanks conclude that exposing people to the name of an object, or to a description of an action, can fool them into thinking they've previously seen a picture of this object or action—causing "striking levels of false recognition."
To produce their results, Weinstein and Shanks had test subjects look at pictures of common objects. The subjects then studied names of objects, some of which they had seen in the pictures, some of which they had not. Ten minutes later the subjects were shown more pictures and had to choose which they recognized from the earlier viewing.
In the end, people accurately identified images they had in fact viewed earlier, and they were also very good at knowing which images they'd never seen at all. Where they failed, write the authors, was in their pronounced tendency to falsely recognize objects whose image was implanted with words but not an actual picture:
It is only when participants are exposed to the name of an object that their ability to reject a new item becomes impaired. ... Interacting with the name of an object led to a significant increase in the number of pictures that were judged remembered ...
So although pictures serve as a useful bridge for memory, it's all too easy to build this bridge out of straw.
Critically, the researchers extended their conclusions beyond mere objects. I wouldn't expect a veteran to recall details of clearing company procedure when shown a simple Red Cross helmet, for instance, but I might expect to elicit vivid memories with a picture of men from his company standing outside the actual clearing station.
Yet in a follow-up trial the researchers discovered action shots—of real people involved in real activities—to be equally corruptible. Simply describing imaginary actions to the test subjects triggered "robust false recognition of photographs depicting people performing those actions."
The authors recognize the real-world implications of their results:
This finding can be readily applied to situations outside the laboratory, including legal contexts, where it would have strong implications in terms of the reliability of recognition memory for physical evidence that had previously been alluded to verbally.
I would add journalistic contexts to this list. Thank you very much, science, for making my week a bit more difficult.
---
* This threat was tested only once, it seems, and the sonuvabitch was indeed whipped.
(HT: BPS Research Digest)
