These experiments make use of the DRM paradigm and investigate whether type of test (Source Monitoring vs. Old/New Recognition) influences false recognition.

In their previous research Hicks and Marsh (1999) found that when the source of items within lists varied false recall was sometimes reduced.  The idea was that if words within the list varied in mode of presentation (e.g. male versus female speaker) that subjects would be more likely to reject the critical lures  (i.e. "If I had really experienced that word I should be able to recall who said it").

By that logic, the effect should be even greater if you emphasize source information by asking subjects to make source judgments.  Hence, Experiments 1, 2A, and 2B, presented in that order and without commercial interruption.

Subjects were presented with DRM lists in which half of the items were presented auditorially and half were derived by the subject from anagrams.

Recognition memory was then tested.  The test included (1) unrelated lures, (2) targets that the subject heard, (3) targets that the subject generated, and (3) critical lures.  Half of the subjects made OLD/NEW judgments and half made HEARD/GENERATED/NEW judgments (i.e. a source monitoring test).

Results

Subjects were pretty accurate on the source test.  They indicated the correct source of the heard items 67% of the time.  They indicated the correct source of the generated items 70% of the time.  They said unrelated lures were new 94% of the time.  For critical lures subjects were more likely to call them heard (57%) than generated (14%).

More critical lures were called "old" under the source monitoring test (71%) than under the recognition test (56%).  This isn't what they predicted.  Signal detection analyses were conducted on the targets (but not the critical lures?).  d' was not influenced by whether a source test was used or an old/new recognition test was used.  However response bias was more liberal with the source test.

Methods

In Experiment 2A they presented some items on each list auditorially and some visually (see Gallo, et al., 2001).  In Experiment 2B a male voice presented some items on each list and a female voice presented other items.

Results

Subjects were quite a bit less accurate in their source attributions in these experiments.  But when it came to source attributions to critical lures they were now more equally divided between the two possible response categories for false memories and that's what Hicks and Marsh were going for.  In Experiment 2A subjects indicated that CLs were seen 32% of the time and that they were heard 48% of the time.  In Experiment 2B subjects indicated that CLs were presented by the male speaker 49% of the time and by the female speaker 41% of the time.

Critical lures were again more likely to be falsely recognized by those taking the source test than by those taking the old/new test. Again, d' was not influenced by whether a source test was used or an old/new recognition test was used.  Response bias was more liberal with the source test.
 

The results didn't turn out that way.  In point of fact, false recognition was higher in the source monitoring condition than in the old/new recognition condition.

One possibility is that the source monitoring test provides subjects with three choices, two of which constitute saying that the item was previously encountered.  The Old/New recognition test provides subjects with two choices,only one of which constitute saying that the item was previously encountered.

Another possibility is that when subjects encounter the critical lure on the test it reminds them of related targets.  The source of the related targets then gets attributed to the critical lure (a process similar to what we've called content borrowing, Lampinen, Neuschatz & Payne, 1999).

Another possibility is that source monitoring instructions require subjects to look vigorously for source specifying information and may cause subjects to accept even weak source information.  This may lead to an overall lowering of the response criterion.

Okay, then to the issue of whether, in the normal course of affairs, source judgments occur postrecognition (i.e. You recognize the item as having been presented and only later figure out where you encountered the item). 

• First off, the authors argue that source judgments involved a more liberal response criterion than the old/new judgments.
• Second, they argue that this point implies that the underlying cognitive mechanisms used to determine if an item is old or new are not the same in source memory paradigms as in old/new recognition paradigms.
• Third, they argue that these first two points cause problems for multinomial models because those models assume that source judgments happen after old/new judgments.