In the past, I've often wondered how journalists pick which studies to write about. The obvious answer is that they pick studies that will get readers or viewers, but given how little their stories correspond with the research they're writing about, it seems to me like they could pick any study and make it sellable. So the question remains, how do they pick the studies they write about? I read an article this weekend that provides an answer: they write about a study when they can get the author to say stupid shit. But before I get to the article, let me tell you a little bit about the study it's ostensibly about. I'll start with some background.
Reconstructive Memory and Source Monitoring
Retrieving something from memory is a reconstructive process. An episode in memory is not stored in one place, but instead, features of the episode are stored all over the brain. Such features might include the sensory input from the episode, our affective response to the episode, our background knowledge or concepts associated with the episode and used to interpret it, where and when the episode occurred, etc. The reconstructive nature of memory retrieval has many implications for memory accuracy. For example, when we encode episodes that are familiar (e.g., eating at McDonald's for the umpteenth time), we tend not to encode specific details about the individual episode, but instead encode it as another instance of a script or schema that is already represented in memory (in the case of eating at McDonald's, as an instance of our McDonald's schema). This means that when we're attempting to reconstruct the episode from memory, we're likely to mistakenly include information from the script or schema that was not present in the particular episode. If, when I go to McDonald's, I usually ask for sweet and sour sauce, but in the particular McDonald's episode I had forgotten to ask for sweet and sour sauce, I'm likely to remember asking for it anyway.
Another way in which the reconstructive nature of memory retrieval can affect memory accuracy concerns source monitoring. We've all had the experience of not being sure whether an episode we remember actually occurred, or if we imagined it or even dreamed it. Deciding which is true is called source monitoring. Did I see an event occur, did someone tell me about it, or did I imagine it? Each time we reconstruct an episode from memory, we have to make such a source judgment. Source monitoring processes involves using the detail, familiarity, vividness, salience, context, and other features of the information we retrieve from memory to determine where it might have come from.
In some cases, particularly cases in which a memory is encoded through an interpretive schema (e.g., the McDonald's case above), source monitoring breaks down, and we mistakenly attribute information that came from in the head (i.e., we imagined it or dreamed it) to external events. The clearest examples of such failures in source monitoring are false memories. In the Deese-Roediger-McDermott paradigm that I've previously described, for example, false memories for conceptual information primed by word lists result from the failure to make accurate source judgments about that information. In other words, people fail to recognize that the information came from inside their heads, when it was primed by words on the list, instead of from the list itself.
Source monitoring has been a hot topic in memory research because of its connection to false memories, which in turn connects it to some important practical issues, most notably recovered memories and eye witness testimony. It has been hypothesized that, in at least some cases of recovered memories, the memories were actually implanted, either in the course of therapy or through encountering descriptions of abuse. In such cases, the implanted memories are mistaken for memories of actual instances that the rememberer experienced because he or she fails to recall the source of the information. In eye witness testimony, there are all sorts of possible causes for breakdowns in source monitoring. For example, a witness who has been asked to identify a perpetrator may be asked to view the photos of individuals many times. Doing so may result in a high level of familiarity for particular faces, and increased familiarity can result in source monitoring errors such that the witness mistakenly believes that he or she saw a face during the crime rather than in a photo1.
Source monitoring errors may also be the cause of some of the symptoms of schizophrenia, such as hallucinations and delusions. Hallucinations and delusions result from an inability to recognize that the hallucination or delusion is imagined, rather than real. Consistent with this, several studies have reported that schizophrenics are particularly bad at determining whether the source of information is the external world or their own imagination in ordinary tasks2. Recently, imaging studies have implicated three brain regions associated with hallucinations in schizophrenics: the medial anterior prefrontal cortex, the thalamus, and the cerebellum3. All three of these regions are also associated with memory in healthy individuals. These facts led Simons et al. (see footnote 3) to hypothesize that the function of these regions in memory might be related to source monitoring.
An fMRI Investigation of Source Monitoring
To test this hypothesis, Simons et al. designed a study in which participants would make source monitoring judgments in an fMRI machine. In one condition, the "perceived condition," participants were presented with highly familiar word pairs (e.g., "Laurel and Hardy" or "Rock and Roll", from p. 2), one at a time, and were asked to count the number of letters in the second word (e.g., "Roll" in the pair "Rock and Roll"). In the second condition, the "imagined condition," participants were presented with the left word from word pairs (again, one at a time), and a question mark in place of the second word, and were asked to imagine the second word and count the letters in it. In both conditions, the words were presented at varying positions on the screen, and each participant saw some of the words in the perceived condition format and some in the imagined condition format.
Immediately following this study phase, the test phase began, and participants were presented with cues consisting of the first words of word pairs they had seen in the study phase. For each cue, participants were asked to either indicate the position on the screen that the word had appeared in the study phase (word-position condition) or to indicate whether the second word in the pair had been perceived or imagined (source-monitoring condition). Each participant received an equal number of both types of questions over the course of the test phase. Participants completed the four blocks, each consisting of a study and test phase, and the experimenters gathered imaging data using functional magnetic resonance imaging (fMRI) during both phases over all four blocks.
In order to determine which regions were associated with source monitoring, Simons et al. contrasted the activation observed during the source-monitoring condition with that observed during the word-position condition. This allowed them to subtract out activation associated with general memory processes, and focus on the areas associated with source-monitoring specifically. The results of this contrast are shown in the figure above (Figure 2, p. 5). As a result of this contrast, they found significant activation in all three areas associated with hallucinations in schizophrenia, including the medial anterior prefrontal cortex (left image), the thalamus (middle image), and the cerebellum (right image), during the source-monitoring condition. The medial anterior PFC, in particular, was associated with source-monitoring errors, as failures to correctly indicate whether the second word in a pair had been perceived or imagined (which occurred 22% of the time) were negatively correlated with activity in this region. So, less activity in the medial anterior PFC resulted in more source-monitoring errors. Simons et al. argue that this is because the anterior PFC is associated with memory for context, and memory for context is important for source-monitoring. They also note that the anterior PFC is associated with the processing of both internal (e.g., goal processing and self-monitoring) and external information. They thus argue that the anterior PFC may play a role in biasing attention "between internally generated and externally derived information" (p. 6). They don't, however, speculate on the roles played by the thalamus and cerebellum. Both regions have previously been associated with memory, including various types of amnesia when there are lesions to these regions, but as far as I can tell, their exact functions in memory are not well understood.
So that's the study. I'm pretty familiar with Simons work on context in memory, and had heard about this study, so I went looking for it on his website (you can read it here). When I read it, I had many thoughts. For example, I thought it cool that we're beginning to understand the neural underpinnings of source monitoring. I also thought that the fact that hallucinations, delusions, and source monitoring may take place in the same brain regions could imply that source monitoring in memory is actually a part of a larger reality monitoring system. One thing I didn't think, though, was that the Simons et al. data said anything about the extent of our grip on reality. Then, while rummaging around at Science Daily, I came across this article. Notice the title: "Our Grip On Reality Is Slim, Says University College London Scientist." Imagine my surprise when I realized it was about the same study I just described to you. How did the author of the article (which is actually the UCL press release) jump from data demonstrating that source monitoring results in increased activation in three brain regions to "our grip on reality is slim?" If you're as consused as I am, then this quote from Simons might help explain the leap:
In our tests volunteers either thought they had imagined words which they had actually been shown or said they had seen words which in fact they had just imagined - in over 20 per cent of cases. That is quite a lot of mistakes to be making, and shows how fallible our memory is - or perhaps, how slim our grip on reality is!
Huh? OK, to be fair to Simons, part of that statement is true: 22% of the time, participants mistakenly indicated that a word had been perceived when it had been imagined, or vice versa. That's more than 20%, just as Simons said. Of course, participants were wrong about the position in which the word had been presented 23% of the time, which, if we're going by the standard in the quote above, means that our grip on location is slightly slimmer than our grip on reality. But does this data really say anything about our grip on reality? Of course not.
The Simons et al. study is not the first study to show that our source monitoring abilities are less than stellar, nor will it be the last. But as I'm sure any philosopher will tell you, reality consists of much more than the source of information, and our grip on reality consists of much more than source monitoring. What this study does show is that after one exposure to an episode in a strange experimental context, our memory for that episode won't be all that great, whether we're trying to remember the source of a part of that episode or where the episode occurred on a computer screen. Frankly, I would be surprised if their memories were extremely accurate under such conditions.
Or course, source monitoring errors do occur in everyday experience (as do errors in memory for locations), and they're probably not all that uncommon. But nothing in the Simons et al. data tells us how often they might occur outside of the laboratory, and thus nothing in this study tells us how slim our grip on reality might be. Why on earth Simons would say such a thing is beyond me. Sometimes, when strange quotes like this from scientists appear in the press, I tell myself that the scientist was probably taken out of context, because I have much more respect for scientists, in general, than I do for journalists. so I like to give the scientist the benefit of the doubt. In this case, however, there's no mistaking the fact that what Simons said about his own study (which is a very interesting study for those of us who study memory) is simply batty. In addition to being batty, though, it's also sexy, so it's not surprising that the press release was picked up by Science Daily (which has never been averse to picking up sexy but batty science stories) and other outlets, and thus inevitably, by the blogs.
1Tredoux, C.G., Meissner, C.A., Malpass, R.S., & Zimmerman, L.A. (2004). Eyewitness identification. In C. Spielberger (Ed.), Encyclopedia of Applied Psychology, V. 1, pp. 875-887, Academic Press.
2E.g., Keefe, R.S.E., Arnold, M.C., Bayen, U.J., McEvoy, J.P., & Wilson, W.H., (2002). Source-monitoring deficits for self-generated stimuli in schizophrenia: multinomial modeling of data from three sources. Schizophria Research, 57, 51-67.
3Simons, J.S., Davis, S.W., Gilbert, S.J., Frith, C.D., & Burgessa, P.W. (In Press). Discriminating imagined from perceived information engages brain areas implicated in schizophrenia. NeuroImage.
I love your blog and thanks for covering this topic! When I saw the title in the RSS feed, I thought you may actually be talking about a study that I was personally involved in and recently published, but you didn't reference it.
For your and other readers' edification on this area, I encourage you to check out our article, which is very related to that of the Simons group and with similar findings:
Hi Brian, thanks for the reference. I hadn't read that, but I'm going to check it out in the next couple days. If I post about it (and since I tend to post about what I'm reading, I very well may), I'll let you know. You can let me know what I got wrong.
"The Simons et al. study is not the first study to show that our source monitoring abilities are less than stellar, nor will it be the last."
"So the question remains, how do they pick the studies they write about? I read an article this weekend that provides an answer: they write about a study when they can get the author to say stupid shit."
Only too true. I am reluctant to believe that is their only motivation though.
I would say that intentionally supplied 'spin' sent to them by organisations with a large public relations department and an eye for expansion, is much more important. I have seen many cases and mention this in my own blog.. Also, there is so much new work all the time that there are hard choices.
This may be splitting hairs, but in the Science Daily article you cite, the "slim grip on reality" quote actually seems to come from Dr Burgess rather than Simons...