Rethinking Familiarity: Remember/Know Judgments in Free Recall

Abstract

Although frequently used with recognition, a few studies have used the Remember/Know procedure with free recall. In each case, participants gave Know judgments to a significant number of recalled items (items that were presumably not remembered on the basis of familiarity). What do these Know judgments mean? We investigated this issue using a source memory/free-recall procedure. For each word that was recalled, participants were asked to (a) make a confidence rating on a 5-point scale, (b) make a Remember/Know judgment, and (c) recollect a source detail. The large majority of both Remember judgments and Know judgments were made with high confidence and high accuracy, but source memory was nevertheless higher for Remember judgments than for Know judgments. These source memory results correspond to what is found using recognition, and they raise the possibility that Know judgments in free recall identify the cue-dependent retrieval of item-only information from an episodic memory search set. In agreement with this idea, we also found that the temporal dynamics of free recall were similar for high-confidence Remember and high-confidence Know judgments (as if both judgments reflected retrieval from the same search set). If Know judgments in free recall do in fact reflect the episodic retrieval of item-only information, it seems reasonable to suppose that the same might be true of high-confidence Know judgments in recognition. If so, then a longstanding debate about the role of the hippocampus in recollection and familiarity may have a natural resolution.

Keywords: Familiarity; Recall; Recognition; Recollection.

Figure 1

Average frequency of Remember/Know judgments as a function of confidence rating in Experiment 1 (error bars represent standard errors).

Figure 2

Average frequency of Remember/Know judgments as a function of confidence rating in Experiment 2 (error bars represent standard errors).

Figure 3

Source accuracy scores (and associated standard errors) of the 45 participants who made at least one correct high-confidence Remember judgment and one correct high-confidence Know judgment in Experiment 2.

Figure 4

Hypothetical data predicted by the generate-recognize account. If the recognize-generate strategy were employed, then correct Remember judgments would not be affected, but correct Know judgments would be. As shown, Know judgments would show a selective increase for correctly recalled words.

Figure 5

Average frequency of Remember/Know judgments as a function of confidence rating in Experiment 3 for correctly recalled words (Panel A: Free Recall Group; Panel B: Forced Recall Group) and for incorrectly recalled words (Panel C: Free Recall Group; Panel D: Forced Recall Group).

Figure 6

Average frequency of Remember/Know judgments as a function of confidence rating in Experiment 4 (error bars represent standard errors).

Figure 7

Hypothetical episodic memory search set created by presenting 24 items on a study list. The ~ symbol represents a non-recoverable item, K represents a recoverable item that is associated with little or no source information, and R represents a recoverable item associated with more substantial source information.

Figure 8

Cumulative recall curves for the Free Recall condition of Experiment 3. Panel A shows the results for all Remember and Know judgments (i.e., collapsed across all levels of confidence), and Panel B shows the results for words recalled with high confidence only.