Dodson, C.S. & Schacter, D.L. (2002). When false recognition meets metacognition: The distinctiveness heuristic

Dodson, C.S. & Schacter, D.L. (2002). When false recognition meets metacognition: The distinctiveness heuristic. Journal of Memory and Language, 46, 782-803.

The distinctiveness heuristic refers to a strategic process in which people use metacognitive information about how vivid they expect their memories to be, to reject items that lack the requisite vividness (Dodson & Schacter, 2001; Schacter, Israel & Racine, 1999). For instance, false memories appear to be less common for pictures than for words (Israel & Schacter, 1997; Schacter, Israel & Racine, 1999). According to the distinctiveness heuristic account, this occurs because people expect their memories for pictures to include vivid visual detail that is typically lacking in the lures.

One finding from the past research is that the distinctiveness heuristic seems to apply only in between subject designs. The authors argue that this may be for one of two reasons:

In a within subjects design distinctiveness is not diagnostic because while memory for some items is vivid and detailed, memory for other items is sparse and less detailed. Thus, distinctiveness cannot be used consistently across the test as a whole.
Its possible that subjects may try to use the distinctiveness heuristic on a list by list basis but may not be able to accurately recall which lists were presented in the distinctive format (e.g. pictures) and which were presented in the less distinctive format (e.g. words).

The four experiments the authors conducted used the repetition lag procedure (Underwood & Freund, 1970). In this procedure subjects study a list of items. They then take a recognition test (either forced choice or old/new). Across the test, studied items are presented only once, but lure items are repeated. At short lags, repeating the lure should decrease the probability of incorrectly selecting it, because only new items are repeated. At long lags, repeating a lure should increase the probability of incorrectly selecting it, because repetition will increase the item's familiarity (Notice the similarity to the study we read by Jones & Atchley, 2002). Experiment 1 Subjects were presented with sixty unrelated items. Subjects were randomly assigned to one of three groups:

Word Condition: All sixty items were visually presented words
100% Distinctive: All sixty items were presented as pictures with an accompanying auditory presentation of the word.
50% Distinctive: Half were presented as words and half were presented as pictures with an accompanying auditory presentation of the word.

The logic here is that in the 50% distinctive condition, the utility of using the distinctiveness heuristic should be reduced. Subjects might well reason, lots of the presented items were presented as words. So there is no reason to expect all presented items to be especially vivid in my memory. Alternatively, subjects might continue to use the distinctiveness heuristic, even though it is not perfectly diagnostic.

The test items were visually presented words and the repetition lag procedure was used to increase the familiarity of lures:

Old items presented only once
New items presented twice at a lag of 4, 12, 24, or 48
Subjects were informed that the new items would repeat

Results

The major results of the experiment are as follows:

When it came to the hit rate, distinctiveness had no overall effect
When it came to the false alarm rate for the first presentation of the new item, distinctiveness had no effect.
When it came to the false alarm rate for the second presentation of the new item, the 100% distinctive condition and 50% distinctive condition produced a somewhat lower false alarm rate.
As expected false alarm rates increased as the lag between repeated presentations of the new item increased.

So a big finding here is that subjects appear to be using the distinctiveness heuristic even though it is not perfectly diagnostic. Experiment 2

Experiment 2 they push this point further by varying the proportion of distinctive items in the list. Lists were made up of either 10% pictures, 25% pictures or only words. So the question is, will people still use the distinctiveness heuristic if very few of the items are actually pictures?

Results

The major results were as follows:

With regard to the targets and the first presentation of the lure, condition had no effect on recognition other than making the response bias more conservative in the 25% distinctive condition.
With regard to the second presentation of the lures, the 25% distinctive condition supressed false memories, but the 10% distinctive condition did not.
As expected, false recognition increased with increasing lag

Major result here was that even when only 25% of the studied items were pictures there was still an effect of distinctiveness. Experiment 3 This is where it gets really cool! The authors wanted to see if subjects expectations would influence the use of the distinctiveness heuristic. In particular, participants in some conditions were told that the memory test would only include items studied as words. If the distinctiveness heuristic is a metacognitive strategy, then this group should not rely on the distintiveness heuristic because they should not expect their memories to be very vivid.

Participants were placed in one of the following conditions:

Word Only Condition: Studied only words
Standard Distinctive: Studied a lists made up of 1/3 pictures and 2/3 words. Targets on the recognition test included both items that had been studied as pictures and items that had been studied as words.
Informed Distinctive: Studied a list made up of 1/3 pictures and 2/3 words. Targets on the recognition test included only items that had been studied as words. Participants correctly told that only items that had been studied as words would be tested (this group should abandon the distintiveness heuristic).
Uinformed Distinctive: Studied a lists made up of 1/3 pictures and 2/3 words. Targets on the recognition test included only items that had been studied as words. Participants incorrectly told that both types of items would be tested (this group should use the distintiveness heuristic).

Notice that the only difference between the informed and uninformed groups is what they believe about the test. The tests themselves are exactly the same. So any difference between the two groups will be due to the subjects metacognitive beliefs.

Repetition of the lure occurred either 24 or 48 items after first presentation. The critical comparison was between the informed and uniformed conditions as they were identical except for the expectations participants were led to have.

Results

Here are the major results:

With regard to the targets and the first presentation of the lure, condition had no effect on recognition other than making the response bias more conservative in the 25% distinctive condition.
With regard to the second presentation of the lure, false recognition was supressed (relative to the word condition) in both the standard distinctive and the uninformed distinctive conditions. False recognition was not supressed in the informed distinctive condition.

Pretty cool finding! The distinctiveness heuristic is used even if none of the test items were studied as pictures, as long as subjects think some of the test items were studied as pictures. Experiment 4 Final experiment is used to rule out the possibility that it is hearing the word, not seeing the picture that is producing the effect. Subjects studied items in one of three conditions.

Word Condition: As before, subjects study a visually presented word
Word + Sound: Subjects study a visually presented word and also hear the word
Picture + Sound: Subjects study a picture and hear the word

Lags between first and second presentation of lures was 24 or 48.

Results

Here are the major results:

With regard to the targets and the first presentation of the lure, condition had no effect on recognition other than making the response bias more conservative in the 25% distinctive condition.
With regard to the second presentation of the lure, false recognition was supressed only in the picture + sound condition, not in the word + sound condition.

So the effects they found in all these experiments are due to the presence of the picture, not to the accompanying sound. Some Points Mentioned in the GD Here are some of the conclusions Dodson and Schacter would like you to draw from these results.

They argue that the reason pictorial presentation reduces false recognition is because subjects are using the distinctiveness heuristic. In other words, when pictures are studied, subjects expect their memories to be vivid and so will reject items that are not sufficiently vivid.
This occurs even when a relatively small proportion (25%) of the items were presented as pictures. This is somewhat odd, in that when only a small proportion of items were studied as pictures, memorial vividness is not an especially diagnostic cue (i.e. Many of your memories will lack vividness).
Interestingly they showed that it is not the actual makeup of the test that matters, but subjects' beliefs about the makeup of the test. When subjects believe that the test includes items that should be memorially vivid, the distinctiveness heuristic kicks in and false memories are suppressed.
One unexplained finding is that evidence for the distinctiveness heuristic did not consistently occur on the lure's first presentation. The authors suggest that this may reflect a floor effect.
Results of the lag manipulation also suggest a recall to reject (i.e. recollection rejection) mechanism (check out our summaries on the topic -- Brainerd & Reyna, 2002; Rotello, et al., 2000; Rotello & Heit, 1999; 2000). False recognition increased as the lag between the first presentation of an item and the second presentation of the item increased.


University of Arkansas	Department of Psychology	Graduate Program in Experimental Psychology
Lampinen Lab	False Memory Reading Group	False Memory Reading Group Fall 2002