Journal Article Summaries
The Importance of Time to Think Back: The Role of Reminding in Retroactive Effects of Memory
Jacob H. Negley Southwestern College
Colleen M. Kelley Florida State University
Larry L. Jacoby Washington University in St. Louis
Change has been described as detrimental for later memory for the original event in research on retroactive interference. Popular accounts of retroactive interference treat learning as the formation of simple associations and explain interference as due to response competition, perhaps along with unlearning or inhibition of the original response. By such accounts, providing additional study time for a changed response in a classic A–B, A–D learning paradigm should increase retroactive interference. In contrast, our experiments show that changing a response produces retroactive facilitation rather than retroactive interference but that outcome requires that the change be detected in the form of a reminding. When reminding does not occur, retroactive interference is observed. Increasing time to study the changed response increases the likelihood of being reminded. Accounts in terms of simple associations cannot explain the importance of reminding. We do so by assuming that being reminded results in a recursive representation that includes both the original and changed response along with the order in which they occurred. We discuss the importance of our results for application as well as for theory.
Keywords: reminding, interference, facilitation, retroactive effects
Suppose people were asked to read two articles presenting claims related to the Affordable Care Act, with some contradictory claims between the two articles. After reading the second article, they are asked to recall the claims presented in the first article. Would increasing the amount of time allowed for reading the second article influence their ability to recall the claims provided by the first article? This corresponds to a question about retroac- tive interference, which was frequently investigated using a paired- associate learning paradigm (e.g., Postman & Underwood, 1973). The typical finding was that presenting pairs with shared cues but changed memory targets across successive lists of word pairs for example, pair AB in the first list followed by changed pair AD in the second list, produced retroactive interference, poorer memory performance compared to that for control pairs that did not share cues across lists. Competition between responses negatively af- fects memory as a function of the relative strength of the responses paired with the same cue, and such response competition is an important mechanism of forgetting in many formal models of
memory (e.g., Raaijmakers & Shiffrin, 1981). In line with a strength prediction, studying second list changed pairs repeatedly in a paired associate paradigm can produce increased retroactive interference relative to studying them a single time (Delprato, 2005; Wichuwut & Martin, 1971; but see Underwood & Lund, 1981). From the perspective of response competition, increasing reading for the second article would be predicted to increase the memory strength of claims presented in it and, thereby, increase retroactive interference due to increased response competition. Alternatively, perhaps allowing greater time to read the second article would make it more likely that one would notice the contradictions between statements made in the two articles. No- ticing contradictions might lead to claims from the first article being more likely to be remembered than they would be had the second article not been read.
Such retroactive facilitation rather than the usual retroactive interference between competing claims was reported by Otero and Kintsch (1992). They found that many students failed to detect contradictions between sentences in text but the few who did so showed facilitation of memory for both the original and the con- tradicting sentences, whereas those who did not detect the contra- dictions recalled one or the other of the contradictory sentences or neither. Noticing contradiction requires that the earlier-read claim be brought to mind by reading the later-made claim, which serves as an implicit repetition of the earlier-made claim that could increase the probability of its later recall. Otero and Kintsch found large individual differences among participants in their experi- ment. Noticing contradictions likely requires sufficient time for the corresponding earlier-read claim to be brought to mind along with an inclination to look back to the earlier-read text.
This article was published Online First August 9, 2018. Jacob H. Negley, Department of Psychology, Southwestern College;
Colleen M. Kelley, Department of Psychology, Florida State University; Larry L. Jacoby, Department of Psychology, Washington University in St. Louis.
This research was supported National Science Foundation Grant Num- ber 1430778 awarded to Colleen M. Kelley.
Correspondence concerning this article should be addressed to Jacob H. Negley, Department of Psychology, Southwestern College, 100 College Street, Winfield, KS 67156. E-mail: Jacob.negley@sckans.edu
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
Journal of Experimental Psychology: Learning, Memory, and Cognition
© 2018 American Psychological Association
2018, Vol. 44, No. 9, 1352–1364 0278-7393/18/$12.00 http://dx.doi.org/10.1037/xlm0000512
1352
Prior investigations have shown that both noticing and recol- lection of change are important for proactive interference, the influence of prior learning on memory for a later event. That research has revealed that noticing change and recollection of change at the time of test transforms proactive interference into proactive facilitation (Jacoby, Wahlheim, & Kelley, 2015; Wahl- heim & Jacoby, 2013). Participants studied pairs of words that changed across lists (e.g., knee-bone in a first list, followed by knee-bend in a second list) as in classic proactive interference experiments that test memory for the changed pair. Importantly, during study participants noted if any List 2 pairs reminded them of the corresponding first list pairs, and later during test for the List 2 responses, they indicated whether they recollected that a partic- ular pair had changed. Overall memory for changed pairs did not differ from memory for control pairs that had no corresponding pair to serve as a source of interference, but that lack of interfer- ence was revealed to be a mixture of effects: There was proactive facilitation for pairs when the second list changed pair reminded participants of the corresponding first list pair if the reminding could be recollected at test, but there was proactive interference for pairs when the changed pair did not trigger a reminding. Interfer- ence was most severe when changed pairs reminded the partici- pants of the first pair, but the reminding could not be recollected at the time of test. Similar results occurred for more naturalistic materials, such as memory for a politician’s most recent positions on topics, when those whose positions changed across debates (Putnam, Wahlheim, & Jacoby, 2014).
How can reminding and recollection of the reminding transform proactive interference into proactive facilitation? Jacoby et al. (2015) used the example of being introduced to a woman and then being reintroduced to her years later after she has married and changed her last name. Noticing that her name has changed would create a memory that embeds the earlier name in the experience of hearing the new name (“Mary Jones, oh, she used to be Mary Smith”). Hintzman (2004, 2011) dubbed such memories “recursive remindings” and used the concept to account for people’s ability to judge the frequency of repetitions of items in a studied list. If one can remember being reminded that one saw a word earlier, one can be sure it occurred twice. In the case of the married woman’s name change, when one wants to later remember her current name, remembering that it changed can cue retrieval of the recursive trace to yield the changed name. Even if her original name comes to mind as a potential intrusion error, the combination of the original name and the recollection that it changed will be partic- ularly effective at cuing the recursive trace. Memory for the recursive reminding contains within it memory for the order of the two pairs, as well as memory for the original response, and so intrusion errors can be avoided (see also Jacoby & Wahlheim, 2013; Jacoby, Wahlheim, & Yonelinas, 2013).
The distinction between proactive and retroactive interference has been critically important for theory regarding the mechanisms of forgetting. Initial work focused on retroactive interference and proposed response competition as the mechanism (McGeoch, 1932). Subsequently, Melton and Irwin (1940) argued for a two factor theory of retroactive interference, where response competi- tion and unlearning of the original response both contributed. Later, after Underwood (1945) demonstrated large effects of pro- active interference on forgetting, those effects were attributed to response competition alone (Postman & Underwood, 1973). The
contrast between proactive and retroactive interference continues to be important in current theories, with competition as a shared mechanism, but retrieval inhibition as a mechanism that uniquely contributes to retroactive interference (Anderson, 2003).
Jacoby et al. (2015) examined the role of noticing change during new learning on retroactive as well as proactive effects of memory. They manipulated the likelihood of noticing changed pairs with instructions to either look back across two lists of word pairs to detect if pairs in a second list had changed from those in a first list, the N-back condition, or to look back for changes only within the second list, the within-list condition. The looking back instructions were partly motivated by the individual differences in noticing change found by Otero and Kintsch (1992). Such individual dif- ferences might reflect differences in the extent to which people look back across experiences, which suggested that manipulating looking-back would have effects on noticing change.
The instruction to look back for change across both lists or only look back within the second list had consequences for both pro- active and retroactive effects of memory. In both cases, increased noticing of between-list changed pairs in the N-back condition facilitated cued recall for changed pairs relative to control pairs presented only once. The facilitation did not emerge for between- list changed pairs in the within-list condition, where participants were focused on looking back only within the second list, and so likely noticed fewer between-list changes. For proactive effects of memory, the facilitation depended on the recollection of change at the time of test as well as the prior noticing of change while learning changed pairs. If people were reminded of the earlier response during study of the changed pair, but failed to recollect the reminding at test, proactive interference was exacerbated com- pared to having never been reminded. Retrieval of the List 1 response when the List 2 change was noticed strengthened the List 1 response, and so increased proactive interference.
A critical difference between the retroactive and the proactive case is that in the proactive case, reminding strengthens the com- petitor (the List 1 response) and, so, produces increased proac- tive interference when change is not recollected. In contrast, in the retroactive case, the List 1 response is the target rather than the competitor and, so, reminding strengthens the target. There- fore, unlike the proactive case, there should not be exacerbated retroactive interference when people are reminded of an earlier response during studied of the changed pair, but fail to recollect the reminding at test. Rather, retroactive facilitation should be ob- served. Jacoby et al. (2015) did not measure recollection of re- minding in their experiment that examined retroactive effects of memory. Perhaps reminding alone is sufficient and recollection of being reminded plays no role in producing retroactive facilitation. Alternatively, recollection of the reminding might increase retro- active facilitation beyond that produced by reminding alone. Re- turning to our example of remembering a name, recollecting the reminding might facilitate memory for the original name as well as that for the changed name, with the facilitation being greater than that which would be produced by reminding alone. Access to the recursive representation produced by reminding might cue mem- ory for the original name when the changed name comes to mind, thereby increasing retroactive facilitation. The recursive represen- tation preserves the order in which the original and changed response occurred as well as the two responses. Consequently, a second means by which recollection reminding could increase
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1353REMINDING IN RETROACTIVE EFFECTS OF MEMORY
retroactive facilitation is via identification of the list from which the potential response originated. Our experiments reported here measured recollection of reminding as well as noticing change in the form of a reminding.
Experiment 1 manipulated the likelihood of being reminded by varying the presentation duration of pairs that changed between a first and second list. To encourage reminding participants were told that some pairs would be changed in List 2, and that noticing when pairs changed would benefit their later recall performance (see also Wahlheim & Jacoby, 2013). Being reminded likely requires time. Consequently, we predicted that presenting pairs at a slower rate in the second list would increase the likelihood of being reminded, and that the greater likelihood of reminding would strengthen the original pair to produce better memory for changed pairs compared to control pairs in the slower presentation condi- tion. In contrast, from a response competition perspective, present- ing pairs at a slower rate would be predicted to strengthen the List 2 response and thereby produce increased retroactive interference just as would repeated presentation of the List 2 response (e.g., Delprato, 2005).
In Experiment 1, while studying List 2, participants monitored pairs for changes from List 1, and when they detected a change, they pressed a key and were prompted for the original List 1 response. The test was cued recall of the List 1 targets, and for each item we also asked if they could recollect the reminding at test by asking whether the response paired with the cue had changed across lists, and if so, to give the List 2 response. We then assessed whether memory for List 1 pairs that triggered a remind- ing was sufficient to produce retroactive facilitation, or whether recollection of the reminding at test increased retroactive facilita- tion beyond that produced by reminding alone.
In Experiment 2, we did not require that participants monitor pairs for changes so that we could assess whether relatively more spontaneous reminding of the List 1 responses during study of changed pairs would also produce retroactive facilitation. We measured recollection of reminding at test as an indirect measure of reminding during List 2 to provide converging evidence for our Experiment 1 results that slower presentation increased the likeli- hood of reminding.
Experiment 1
Experiment 1 tested whether longer presentation duration of changed pairs on List 2 leads to a greater probability of reminding than shorter presentation duration changed pairs and so produces retroactive facilitation. After presentation of each pair for study in List 2, participants were asked to report whether the pair changed from List 1, and if so, to report the original response. Successful report of the original response counts as a reminding. We predict a higher rate of reminding for long presentation pairs than for short presentation changed pairs. Additionally, we predict overall retro- active facilitation for List 1 cued recall produced by longer versus shorter presentation duration of changed pairs in List 2. We then test whether reminding is sufficient for retroactive facilitation, or whether recollection of the reminding at test is necessary, as in the case of proactive facilitation. As found by Jacoby et al. (2015), we expect to find retroactive interference for pairs that did not trigger a reminding during the presentation of List 2.
Method
Experiments were approved by Florida State University’s Insti- tutional Review Board. All research was conducted in accordance with current American Psychological Association standards, par- ticularly as they pertain to the protection of human participants.
Participants. Thirty undergraduate students at Florida State University participated in exchange for partial course credit. All participants were tested individually.
Design and materials. The within-participant design varied the relation between List 1 and List 2 pairs, with the latter con- sisting of repetition of pairs from List 1 and interference pairs that were presented for either a relatively short (1 s) or long (7 s in duration). Control pairs in List 1 had no corresponding repetition or interfering pairs in List 2.
All materials were drawn from Wahlheim and Jacoby (2013) and consisted of word pairs with a weak associate relationship between cues and targets (forward, M � 0.03, SD � 0.08; back- ward, M � 0.02, SD � 0.05), as indexed by Nelson, McEvoy, and Schreiber (1998). All critical pairs were created such that for a given cue, there were two possible targets. Targets were ortho- graphically related because they were originally designed to create fragments that could be completed by either of the two responses (e.g., b_n_ could be completed as bone or bend), though fragments were not used as cues in this experiment. Five sets of 20 items were created, with each set matched on length and frequency of cues and targets. Sets were rotated through each within participant item type across participants, resulting in five counterbalances. An additional 12 pairs were used for primacy and recency buffers, and such buffer pairs were not counterbalanced or tested.
List 1 consisted of a total of 106 word pairs (e.g., knee– bone; apple–worm): 100 critical pairs, three primacy buffer pairs, and three recency buffer pairs. The five sets of List 1 pairs took on a different status as a function of what corresponding pairs were presented during List 2, and for what duration: Control pairs for which no corresponding pairs were presented in List 2, interfer- ence pairs for which changed pairs were presented for a short duration or long duration in List 2, and repetition pairs for which the repetitions were presented for a short or long duration in List 2. List 2 consisted of 86 word pairs: 80 critical items, three primacy items, and three recency items. Of the critical items, half were repetitions from List 1 and half were changed, interference pairs, with 20 of each presented for either a short or long duration. At test, participants were presented with all critical cues from List 1. Primacy buffer items and recency buffer items were not in- cluded in the test list.
Procedure. Participants were instructed to study List 1 for a later memory test. Participants were informed that at test they would be presented with the word on the left and asked to recall the word that had been paired with it. List 1 items were presented in a random order for 7 s each followed by a 500-ms interstimulus interval.
Prior to List 2, participants were instructed to study the second list of pairs for a later memory test. Participants were also told that some word pairs would have a different target than the one presented in the first list and that noting this change would help them remember both responses on the memory test later, as in the instructions used by Wahlheim and Jacoby (2013) Pairs were presented in a random order, with half presented at the short
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1354 NEGLEY, KELLEY, AND JACOBY
duration of 1 s and half presented at the longer duration of 7 s, with a 500-ms interval between pairs.
While studying List 2, participants were instructed to explicitly respond when they detected a changed pair. After studying a pair, a screen appeared with the text, “Did second word change from List 1 to List 2?” with a box labeled “yes” and a box labeled “no.” Participants had 5 s to click a response box using the mouse cursor. If participants failed to respond to the change detection question within the 5 s, the program presented the next pair for study. If the participant clicked the “yes” box, a new screen appeared with the question, “What was the word on List 1?” and the experimenter recorded the participant’s response. After recording the partici- pant’s response, the participant clicked the “no” box on the pre- vious screen to move on to the next to-be-studied pair.
At test, after attempting to recall the List 1 response, participants were asked if they recollected whether a pair had changed between Lists. Participants were given six options ranging from “I remem- ber it changed,” “It likely changed,” “It may have changed,” to “It may not have changed,” “It likely did not change,” and “I remem- ber it did not change.” If a participant chose one of the first three options, a new screen appeared with the test “What was the word on List 2?” and the experimenter recorded the participant’s re- sponse. After recording the response, or if the participant chose one of the latter three options, the cue was presented for recall of the next pair.
Results and Discussion
Correct recall and intrusion errors. The probability of cor- rect recall of the List 1 target and of intrusion errors of the List 2 target is reported in Table 1. Not surprisingly, repetitions of a pair in List 2 improved memory relative to control pairs presented a single time in List 1, for both long, t(29) � �10.78, p � .001, Cohen’s d � �1.97, and short duration pairs t(29) � �10.09, p � .001, Cohen’s d � �1.85. The two did not differ, t(29) � �1.02, p � .32, which may be a matter of power, but which is not the focus of our experiments.
As predicted, we found retroactive facilitation in cued recall of change pairs, particularly when the potentially interfering List 2 changed pairs were presented more slowly. Presentation of changed pairs in List 2 improved memory of the corresponding List 1 target relative to List 1 control pairs when the changed pair was presented for the long duration, t(29) � 2.09, p � .05, Cohen’s d � .38 but not when the changed pair was presented for
the short duration, t(29) � .33, p � .75. The two did not differ, t(29) � 1.54, p � .14, Cohen’s d � .12.
The probability of producing the response from the changed pair from List 2 as an intrusion error was much higher for List 1 change pairs than control pairs. Intrusions were significantly more likely for long presentation change pairs as compared to control pairs, t(29) � �8.35, p � .001, Cohen’s d � �2.12, and for short presentation change pairs as compared to control pairs t(29) � �7.19, p � .001, Cohen’s d � �1.65, and the probability of an intrusion error did not differ between long short presentation of changed pairs, t � 1. The intrusion errors indicate the presence of interference in the form of response competition for pairs that change between lists.
Change detection and reminding. We predicted that the retroactive facilitation when corresponding changed pairs in List 2 were presented for a long duration would occur if there were a greater probability of reminding during List 2 for long duration changed pairs. As predicted, detection that a pair had changed in List 2 occurred more often for long duration pairs (M � .59, SE � .03) than for short duration pairs (M � .49), t(29) � 3.42, p � .002, Cohen’s d � .64. Critically, change detection was accompa- nied by memory of the List 1 response, that is participants were reminded of the List 1 response, more frequently for long presen- tation pairs (M � .44, SE � .03) than for short presentation pairs (M � .36, SE � .03), t(29) � 2.80, p � .01, Cohen’s d � �.51. The presentation duration manipulation affected the probability of change detection as well as the probability of reminding. This difference occurred despite the fact that participants had 5 s to enter their online change detection decision by using the mouse cursor to click the “yes” or “no” button after the “did it change?” question appeared, which could have effectively increased the time to detect change in the short presentation condition from 1 s to 6 s, and for the long duration presentation condition from 7 s to 12 s. However, they did not appear to use the change detection response time to attempt to detect change. As shown in Table 2, change detection responses were made quickly. Trials where par- ticipants did not make a change decision before timing out were averaged as the full 5 s. Trials that timed out were relatively rare, 3.2% of all trials and 5.2% of the trials for short duration changed pairs. To the extent that participants did use the time allowed to enter the change detection response to engage in change detection would actually mitigate against our finding of a difference in reminding between short and long duration changed pairs, as it
Table 1 List 1 Mean Recall and Mean Intrusion of List 2 (L2) Response as a Function of Item Type in Experiment 1 and Experiment 2
Corresponding List 2 presentation duration
Item type
A–B, A–B Control A–B, A–D
Experiment Response Long Short None Long Short
Experiment 1 Correct .72 (.03) .70 (.04) .41 (.03) .47 (.03) .42 (.04) L2 intrusion .02 (.01) .01 (.01) .01 (.01) .22 (.03) .21 (.03)
Experiment 2 Correct .71 (.04) .67 (.03) .45 (.04) .55 (.03) .49 (.04) L2 intrusion .01 (.01) .02 (.01) .02 (.01) .15 (.01) .12 (.01)
Note. SEMs are presented in parentheses.
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1355REMINDING IN RETROACTIVE EFFECTS OF MEMORY
reduces the relative difference of our presentation rate manipula- tion.
Reminding as a source of retroactive facilitation. The find- ing that List 2 intrusion errors were higher for change pairs than for control pairs suggests that recall performance reflects a com- bination of retroactive interference produced by competition and retroactive facilitation produced by reminding during study of the second list. Table 3 shows recall conditionalized on whether par- ticipants were reminded during study of List 2 pairs, and reveals that reminding was necessary for retroactive facilitation to be observed at test. When participants were reminded of the List 1 response while studying a List 2 changed pair, ability to recall the List 1 response at the time of test was near perfect, showing massive retroactive facilitation. In contrast, when change was detected but not accompanied by the List 1 response or when change was not detected, massive retroactive interference was observed.
To reveal the degree to which being reminded of the List 1 response while studying List 2 produces retroactive facilitation and failures to be reminded produces retroactive interference, we ex- amined List 1 recall contingent on reminding during List 2 encod- ing, detection of change without reminding, or failure to detect change. Differing degrees of freedom in contingency analyses are due to exclusion of participants who did not have one observation per cell. Reminding of the List 1 response produced subsequent facilitation for corresponding List 1 recall relative to control pairs both for longer, t(22) � 11.99, p � .001, Cohen’s d � 2.53, and shorter presentation duration changed pairs, t(22) � 11.30, p � .001, Cohen’s d � 2.37. Failure to detect change, indicated by a “no” to the question of whether a pair in List 2 had changed, resulted in significant interference compared to control pairs for corresponding List 1 pairs for both longer, t(22) � �7.42, p � .001, Cohen’s d � �1.08, and for shorter presentation duration
change pairs, t(22) � �7.24, p � .001, Cohen’s d � �1.51. Importantly, noticing change without being reminded of the orig- inal List 1 response also produced significant interference for recall of the List 1 change pairs, both when longer, t(22) � �6.28, p � .001, Cohen’s d � �1.28, and shorter presentation duration changed pairs were presented on List 2, t(22) � �5.41, p � .001, Cohen’s d � �1.15. To escape retroactive interference and dem- onstrate retroactive facilitation, one has to be reminded of the List 1 response while studying the changed pair. Noticing that a pair changed without being reminded of the List 1 response is not sufficient. The conditionalized results fit with a dual process model of retroactive effects as a mixture of facilitation due to reminding and interference due to response competition in the absence of reminding.
It is noteworthy that reminding produced a probability of sub- sequent recall that is substantially higher than produced by repe- tition of the List 1 pair in List 2 (.94 vs. .71). That result converges with findings that retrieval can have a larger effect on later mem- ory performance than does restudy, particularly when test condi- tions are made more difficult by retroactive interference or delay (Halamish & Bjork, 2011). Part of the advantage for remindings over repetitions may be due to item selection effects at play when recall is conditionalized on reminding because a List 1 pair has to be sufficiently well remembered to support a reminding during List 2. Weakly remembered pairs are effectively filtered out when test performance is conditionalized on prior reminding whereas they are not when looking at the effect of repetition. Item effects also likely play a role in generating interference: If the List 1 response was not remembered well enough to be spontaneously retrieved during List 2 in the form of reminding, it may not be able to be deliberately retrieved on the cued-recall test.
Is recollection of the reminding necessary for retroactive facilitation? In the case of proactive interference, proactive fa- cilitation depends on being able to access the recursive represen- tation produced by a reminding (Wahlheim & Jacoby, 2013) so as to recall the List 2 target, as outlined in the introduction. As a test of whether the reminding of the List 1 response during study of List 2 is sufficient for retroactive facilitation, or whether partici- pants must additionally gain access to the recursive reminding at test, as is the case for proactive interference (Wahlheim & Jacoby, 2013) we conditionalized List 1 recall on the presence or absence of a reminding during List 2, and on the presence or absence of recollection of the reminding, that is, recall of the List 2 changed response at test. As shown in Table 4 when a reminding occurred and participants could recall the reminding at test, recall of the List
Table 2 Mean Reaction Time for Change Detection Decision During List 2 by Item Type and Duration
Item type Reaction time (ms)
Repetition (long) 927.0 (305.5) Repetition (short) 1,339.1 (481.1) Interference (long) 1,099.88 (376.3) Interference (short) 1,779.1 (558.5)
Note. SDs are presented in parentheses.
Table 3 Recall of List 1 Change Pairs at Test Contingent on Whether Participants Were Reminded of the List 1 Target by the Changed Pair in List 2, Simply Noticed That the Pair Changed in List 2, or Did Not Notice the Change
Corresponding List 2 presentation duration
Item type
Reminded “Yes” change
without reminding “No” change Control
Long Short Long Short Long Short None
Experiment 1 .94 (.02) .94 (.04) .10 (.03) .13 (.05) .11 (.03) .14 (.04) .41 (.04)
Note. SEMs are presented in parentheses.
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1356 NEGLEY, KELLEY, AND JACOBY
1 response was nearly perfect. A 2 (Recollection of Reminding vs. No Recollection of Reminding) � 2 (Duration of List 2 Pair Long vs. Duration of List 2 Pair Short) analysis of variance revealed a main effect of whether the reminding was recollected or not, F(1, 21) � 7.97, p � .01, �p2 � .28, a nonsignificant effect of duration of the List 2 pair, F(1, 21) � 2.79, p � .11, �p2 � .12, and no interaction, F(1, 21) � 1.39, p � .25, �p2 � .06.
It appears that the strengthening effect of the reminding of a List 1 response during the study of the changed pair in List 2 is primarily responsible for retroactive facilitation, and failure to access the recursive reminding at test does not produce the espe- cially severe interference that is observed in the case of proactive interference. In fact, there was a benefit for recall performance when there was a reminding even if it were followed by a complete lack of memory at test for the reminding, as indicated by judging at test that a change pair had not changed between lists. There were few observations of this sort, yet a memory benefit emerged compared to control pairs (Mbenefit � .30), t(24) � 3.99, p � .001. Although it is not necessary to remember the reminding at test to gain a benefit, there is an additional benefit to accessing the recursive reminding at test, because it contains within it memory of the order of the two responses that can aid in list discrimination as in Jacoby et al. (2013). Because these conditionalized analyses are subject to item selection effects, we also address the issue with regression analyses below.
Linear mixed-effects analysis. Although we found more re- minding when changed pairs were presented for a long duration, and retroactive facilitation in overall cued recall, the analyses conditionalized on reminding and on recollection of reminding at test are susceptible to item and subject selection effects. To address these concerns, we conducted a linear mixed-effect analysis that allowed us to examine the effects of reminding and recollection of reminding on the likelihood of recall while including participants and items as random effects (Bates, Mächler, Bolker, & Walker, 2015; R Core Team, 2013). The models were binomial logistic regressions, and p values for random effects were calculated using the likelihood ratio tests comparing the full model against a model without the effect in question. The final model allows random intercepts for both participants and items; adding random slopes caused the models to fail to converge. As is evident in Table 5, Item Type (short vs. long presentation of competitor in List 2) did not influence recall of change pairs. Importantly, reminding had a significant effect on recall performance, reflecting the pattern that reminding during List 2 improved memory for the corresponding List 1 pair even with participants and items as random effects. In
accord with the analysis of conditionalized results, recollection of reminding also had a significant effect on recall performance, reflecting a benefit of being able to remember a reminding at the time of test.
The important difference between proactive and retroactive fa- cilitation is that the target response is strengthened due to retrieval when being reminded in the case of retroactive effects whereas it is the competitor that is strengthened when being reminded in the case of proactive effects. The opposite effects of recollection of reminding and strengthening of the competitor in the proactive interference case led to considerable variance in List 2 recall being explained by individual differences in recollection of reminding (Wahlheim & Jacoby, 2013). In contrast, strengthening effects of the first list response during a reminding and the order information contained in the recollection of a reminding (Jacoby et al., 2013) both serve to increase retroactive facilitation. Recollection of re- mindings contributes to retroactive facilitation, likely because rec- ollecting the reminding at test provides useful information about which target came first.
The current experiment varied presentation duration of the changed pairs in List 2 in a contrast that pitted greater response competition against a higher likelihood of reminding for the pairs presented more slowly. A reminding is a form of spontaneous memory retrieval (Benjamin & Ross, 2011), and some have argued that spontaneous retrieval is relatively fast (Schlagman & Kvavilashvili, 2008) while others have found that spontaneous retrieval is slightly slower than intentional retrieval (Kompus, Eichele, Hugdahl, & Nyberg, 2011). One could argue that the requirement for participants to monitor whether pairs had changed and to report the original pair pushed the task toward more inten- tional retrieval, and that spontaneous reminding would not play
Table 4 Recall of List 1 Change Pairs Contingent on Both Reminding During List 2 and Recollection of the Reminding at Test as Indexed by Correct Report of What Target Changed to in List 2
Corresponding List 2 Presentation duration
Item type
Reminding � recollection of
reminding at test
Reminding � no recollection of
reminding at test No reminding
Long Short Long Short Long Short
Experiment 1 .98 (.02) .99 (.01) .88 (.05) .96 (.02) .11 (.03) .16 (.04)
Note. SEMs are presented in parentheses.
Table 5 Final Model for Linear Mixed-Effect Analyss From Experiment 1
Coefficient Estimate (b) SD SE b �2 z df p
Fixed effects Intercept �2.77 .25 �10.99 .001 Item duration in List 2 .29 .22 1.29 .20 Reminding 5.26 .33 15.85 .001 Recollection of
reminding 1.79 .33 5.48 .001 Random effects
Items .89 12.65 1 .001 Subjects .45 3.30 1 .07
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1357REMINDING IN RETROACTIVE EFFECTS OF MEMORY
such a role in traditional studies of interference. In Experiment 2, we drop the requirement that participants monitor change and report reminding to see if reminding nonetheless continues to occur spontaneously while studying changed pairs. We use recol- lection of the reminding at test as an indirect measure that a reminding occurred during study of the second list. We predict that even under these conditions, participants will be more likely to notice changed pairs and be reminded of the original response when changed pairs are presented more slowly, and so show retroactive facilitation in cued recall.
Experiment 2
Method
Participants. Thirty undergraduate students at Florida State University participated in exchange for partial course credit, and all participants were tested individually.
Design, materials, and procedures. The design and materials for Experiment 2 are as in Experiment 1, but without the measure of change detection and reminding during study of List 2, and a change in the wording of the recollection of change measure. At the beginning of List 2, participants were informed that some pairs would be repetitions, and some would be changed, and that notic- ing that pairs had changed would benefit their memory. Pairs were presented in a random order, with half presented at the short duration of 1 s and half presented at the longer duration of 7 s, with a 500-ms interval between pairs.
At test, the left member of each List 1 pair was presented as a cue to recall the right member of the pair presented in List 1. Recollection of reminding was measured after recall of each item. A screen would appear with the text “Did second word change from List 1 to List 2?” with a box labeled “yes” and a box labeled “no.” If the participant clicked the “yes” box, a new screen would appear with the test “What was the word on List 2?” and the experimenter recorded the participant’s response. After recording the participant’s response, or the participant clicking the “no” box on the previous screen, the cue for the next pair would appear.
Results and Discussion
Correct recall and intrusion errors. Probability of correct recall (A–B) and intrusion errors (A–D) are reported in Table 1 (Experiment 2) for each combination of conditions. As in Exper- iment 1, repeating pairs from List 1 during List 2 led to better recall compared to control pairs presented only in the first list, for both long, t(29) � 8.99, p � .001, Cohen’s d � 1.66, and short,
t(29) � 7.41, p � .001, Cohen’s d � 1.37, repetition items. As in Experiment 1, long and short repetitions did not differ, t(29) � �1.61, p � .12.
As in Experiment 1, we found retroactive facilitation in cued recall of change pairs, but only when the potentially interfering changed pairs in List 2 were presented more slowly. Recall was significantly higher for List 1 change pairs for which there were long presentations of the corresponding changed pairs in List 2 as compared to List 1 control pairs, t(29) � �3.53, p � .001, Cohen’s d � �.65. In contrast, recall was not significantly higher for List 1 interference pairs for which there was a short presenta- tion of the corresponding changed pairs in List 2 as compared to control items, t(29), � 1.45, p � .16. The difference in recall for the List 1 pairs where the corresponding List 2 changed pair was presented for a long versus short duration did not reach the conventional level of significance, t(29) � 2.044, p � .05, Cohen’s d � .37, although it was similar to the pattern in Experiment 1. Retroactive facilitation when the changed pairs had been presented for a longer duration pair but not when the changed pairs had been presented for a shorter duration supports the prediction that longer presentation rates on List 2 increased the likelihood that partici- pants noticed changed pairs across lists and were reminded of the original response, thus strengthening the original response.
Intrusions errors were significantly more likely for long presen- tation change pairs as compared to control pairs, t(29) � �9.12, p � .001, Cohen’s d � �1.96, and for short presentation change pairs, t(29) � �7.95, p � .001, Cohen’s d � �1.60, indicating the presence of interference and response competition from changed pairs in List 2. As in Experiment 1, the probability of an intrusion error did not differ between long and short presentation change pairs, t(29) � �1.38, p � .18. It is possible that an increase in competition when List 2 changed pairs are presented for a longer duration is partially offset by a greater likelihood of reminding for longer duration List 2 changed pairs, which we assess in the following analyses of recollection of reminding.
Recollection of the reminding. The likelihood of reminding during List 2 presentation of changed pairs can be inferred from participants’ recollection of the reminding at test, defined as re- sponding “yes” to the question, “did it change?” and the ability to report the List 2 response. As shown in Table 6, recollection of remindings was greater when changed pairs had been presented for a longer rather than shorter duration, in line with the greater remindings revealed by the direct monitoring measure used in Experiment 1. For comparison, rates of recollection of remindings from Experiment 1 are also presented in Table 6, and are quite similar. Recollection of reminding was higher when the changed
Table 6 Probability of Recollection of Reminding at Test for Change Pairs, and Recollection of Change Without Memory of List 2 Response
Corresponding List 2 presentation duration
Item type
Reminding Recollect change No memory of change False alarm
“changed” to control
Long Short Long Short Long Short None
Experiment 1 .27 (.03) .19 (.02) .24 (.03) .25 (.03) .49 (.03) .56 (.03) .13 (.02) Experiment 2 .30 (.04) .18 (.02) .21 (.02) .20 (.02) .49 (.04) .62 (.03) .07 (.01)
Note. SEMs are presented in parentheses.
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1358 NEGLEY, KELLEY, AND JACOBY
pair had been presented for 7 s compared to 1 s (M � .30, SE � .04 vs. M � .18, SE � .03), t(29) � 3.75, p � .001, Cohen’s d � .78. The slower duration presentation of changed pairs increased the likelihood of reminding even under the more spontaneous conditions of the current experiment.
Recall conditionalized on recollection of reminding. We analyzed List 1 recall of change pairs contingent on recollection of reminding, versus on judgments the pair had changed or had not changed between lists (see Table 7 for means of conditionalized recall and Table 6 for frequency of responses upon which the conditionalized results were computed). Recollection of reminding (with recall of the List 2 response) was associated with significant retroactive facilitation for List 1 change pairs as compared to control pairs both for long duration presentation of the changed pairs in List 2, t(24) � 10.67, p � .001, Cohen’s d � 2.36, and shorter duration presentations, t(24) � 7.92, p � .001, Cohen’s d � 1.59. Recollecting that a pair had changed without the ability to remember the changed response also produced retroactive fa- cilitation of List 1 change pairs relative to control pairs, for both the longer duration presentation pairs, t(24) � 2.64, p � .014, Cohen’s d � .53 and the shorter duration presentation pairs, t(24) � 2.88, p � .008, Cohen’s d � .81. However, in both cases, retroactive facilitation of List 1 change pairs was greater when the List 2 response was remembered than when it was not, t(24) � 4.50, p � .001, Cohen’s d � 1.08 for longer duration and t(24) � 3.26, p � .003, Cohen’s d � .69 for shorter duration List 2 presentations. Based on the results of Experiment 1, cases where people claim to remember that a pair had changed, but cannot remember what the pair changed to are likely a combination of remindings that occurred but that are now partially forgotten (which would facilitate recall); change that was noticed without being reminded of the original response (which would produce interference); and false claims of remembering change that was never noticed (which would produce interference). False recollec- tion of change occurred for control pairs at a rate of .07. Recol- lection of a reminding is extremely unlikely to be a false alarm. In line with the suggestion that recollection of a reminding relies on access to a recursive trace, it essentially never occurred that participants intruded the List 2 response as being the List 1 response followed by reporting the List 1 response as having been the List 2 response (.006).
Retroactive interference only emerged when participants failed to recollect change, with worse memory for change pairs as com- pared to control pairs for both longer, t(24) � 3.95, p � .001,
Cohen’s d � 0.79, and shorter presentation durations, t(24) � 4.06, p � .001, Cohen’s d � 0.81, of the changed pairs in List 2. We infer from the conditionalized recall that when participants were reminded of the List 1 response during study of a changed pair List 2, they later experienced retroactive facilitation in their memory for the List 1 responses and when they failed to be reminded during List 2, they later experienced retroactive inter- ference in their memory for the List 1 responses. The fact that participants were more likely to be reminded during longer pre- sentations of changed pairs on List 2 than during the shorter presentations on List 2 produces the significant facilitation of long presentation change pairs in overall List 1 recall, relative to control pairs.
In sum, the overall probability of recall in a traditional “inter- ference condition” without requiring participants to monitor pairs for change and report the original response reflects a combination of the facilitating effects of reminding on memory for the original responses, and the negative effects of response competition when reminding fails. Depending on the likelihood of reminding, overall recall can tilt toward facilitation when there are more cases of reminding, as in the case for long duration A–D pairs, or the two opposing processes can offset each other, as in the case for the short duration A–D pairs. The contingency analyses based on recollection of reminding revealed that the lack of interference in overall recall for short duration change pairs is due to the level of reminding in the short duration condition and the substantial facilitative effects of reminding.
General Discussion
In two experiments we found that additional study time of changed pairs resulted in a greater likelihood of being reminded of the original response. Longer presentation rates resulted in greater change detection and a higher likelihood of retrieving the original responsse while studying the List 2 changed pairs, which we refer to as remindings (Experiment 1) and greater recollection of re- minding (Experiments 1 and 2). The consequence of being re- minded of the original response during study of the List 2 changed pair was retroactive memory facilitation rather than the traditional memory interference for change pairs, whereas failure to be re- minded of the earlier pairs by the changed pairs resulted in retro- active memory interference (Jacoby et al., 2015). A linear mixed effects analysis provided evidence that these effects were not due to item differences or participant differences (Experiment 1). The
Table 7 List 1 Recall of Change Pairs Contingent on Recollection of the Reminding as Indexed by Report of the List 2 Response at Test, Recollection of the Change Without Memory for What the Pair Changed to on List 2, or No Memory for Change
Corresponding List 2 presentation duration
Item type
Recollect reminding Recollect change Control No memory for
change
Long Short Long Short None Long Short
Experiment 1 .77 (.07) .89 (.03) .63 (.06) .68 (.05) .44 (.03) .20 (.05) .22 (.04) Experiment 2 .94 (.02) .90 (.04) .69 (.06) .66 (.07) .48 (.04) .27 (.04) .32 (.04)
Note. SEMs are presented in parentheses.
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1359REMINDING IN RETROACTIVE EFFECTS OF MEMORY
results support a dual-process model of retroactive interference such that reminding produces retroactive facilitation by strength- ening the original response via retrieval. When reminding does not occur, response competition produces retroactive interference. In contrast to work on proactive facilitation (Wahlheim & Jacoby, 2013), a reminding does not have to be later accompanied by recollection of the reminding at test to produce retroactive facili- tation. In what follows we consider the importance of our results for both theory and for applied concerns.
Implications for Theories of Interference
Two-factor interference theory (Melton & Irwin, 1940; Postman & Underwood, 1973) held that retroactive interference differs from proactive interference in that retroactive interference is caused by unlearning and response competition whereas proactive interfer- ence is caused only by response competition. More modern theo- ries, such as Mensink and Raaijmakers (1988) can account for many classic interference results via response competition, con- textual fluctuation across time, along with the storage of context with pairs presented in different lists. People use contextual cues to accomplish list differentiation when particular responses are re- trieved. The focus of those theories has been on interference effects whereas we propose a dual-process model that accounts for both facilitation and interference effects, and points toward impor- tant differences between proactive and retroactive effects that have been ignored by prior theorizing. We have shown the importance of incorporating remindings as an additional mechanism to under- stand retroactive effects as well as proactive effects in memory. For proactive facilitation, both reminding and recollection of change at test are required because reminding alone increases the strength of the competitor and thereby produces proactive inter- ference (Jacoby et al., 2015; Wahlheim & Jacoby, 2013). In contrast, reminding alone is sufficient to produce facilitation in the retroactive case, and in that case the forgetting of the reminding at test does not counteract the facilitation produced by the reminding. We did find a benefit of recollection of the reminding in the conditionalized analyses of Experiment 1 but even a reminding that was followed by failure to recollect the reminding at test was accompanied by retroactive facilitation for the List 1 response. The logistic regressions that controlled for item and participant selec- tion effects also found an additional benefit for recollection of reminding at test beyond that produced by the reminding.
A recursive representation that results from a reminding in- cludes both the List 1 and List 2 response along with the order in which the occurred. Consequently, reliance on the recursive trace for accessing a response at the time of test would produce positive correlation between recall of the two responses, rather than show- ing an inverse correlation as predicted by response competition and unlearning accounts. Against this backdrop, it is surprising that supposedly competing List 1 and List 2 responses are often recalled independently of one another (Wichuwut & Martin, 1971; Tulving & Watkins, 1974). Hintzman (1972) pointed out that item and subject selection effects would be a source of positive depen- dence that would off-set the negative dependence between re- sponses created by competition. Recently, Burton, Lek, and Ca- plan (2017) found that even when subject-variability effects were controlled for, there was overall independence between recall of the two responses (using the modified-modified free recall
[MMFR] test procedure) but that there was a range of individual- participant correlation values, from very negative to very positive, suggesting variations in strategy could change the relation. In a second experiment, they instructed participants to separately en- code the two lists or to integrate the encoding of corresponding changed pairs while studying the second list. The integrative image condition produced a robust positive correlation between recall of the two responses. Integrating the encoding of changed pairs requires that the List 2 pair remind participants of the List 1 pair, and likely creates an integrative recursive representation. Recol- lection of remindings at test would contribute to positive depen- dency. Our test procedure does not permit us to assess the relation between recall of the two responses, because if participants did not remember that the response had changed between lists, they were not given the opportunity to report a second list response. How- ever, individual variations in the likelihood of reminding could lead to corresponding variation in relation between recall of first and second list responses as seen by Burton et al.
We hold that reliance on recollection of reminding can serve as a means of accessing a sought after response and, thereby produce both proactive and retroactive facilitation. In contrast, others have focused on postaccess identification of list membership when theorizing about interference effects (e.g., Winograd, 1968). Iden- tification of the list membership of a potential response would allow one to avoid intrusion errors such as producing a List 2 response as being a List 1 response. The argument for the benefit of relying on postaccess identification is the same as that made for monitoring source memory (e.g., Johnson, Hashtroudi, & Lindsay, 1993). Source monitoring can serve as a means of avoiding errors and if sufficiently successful could fully eliminate interference effects, producing performance that did not differ from an A–B, C–D control condition. However, postaccess rejection of compet- ing responses could not produce levels of performance that exceed that of the control condition, that is, proactive or retroactive facilitation. In this vein, we have argued for the importance of distinguishing between early selection (means of memory access) and late correction (postaccess monitoring) as forms of cognitive control (e.g., Jacoby, Kelley, & McElree, 1999).
Current models of interference effects (Mensink & Raaijmakers, 1988; Lohnas, Polyn, & Kahana, 2015) have argued that the context in which an item occurs is included in its encoding and that context can be used as a cue for access to the item as well as a postaccess means of avoiding intrusions. If context is sufficiently different for the original and changed pair, interference effects could be avoided by using context as a cue to access the sought after response. However, a problem for this approach is that reminding would be expected to result in the original response occurring in the same context as the changed response, increasing the similarity of the contexts with the result that it would be more difficult to selectively access the target response (for a similar approach regarding the consequences of being reminded of the prior list context on later intrusions see Sederberg, Gershman, Polyn, & Norman, 2011). Also, the implication seems to be one of independence between access to the original and the changed response (cf., Martin, 1971). As described above, independence of responses does not always occur. Further, independent access to the responses would not account for the importance of recollection of reminding for producing proactive and retroactive facilitation. Where other models focus on relatively simple associations and
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1360 NEGLEY, KELLEY, AND JACOBY
competition to account for retroactive and proactive interference, our dual process model and the evidence for it demonstrate the important effects of the recursive representations that are estab- lished by reminding and the recollection of reminding.
Spontaneous Versus Directed Retrieval
Remindings are defined by Hintzman (2004, 2011), and by Benjamin and Ross (2011) as due to spontaneous rather than directed retrieval, and differences between the two are important to identify. The remindings in our experiment when change detection and reporting remindings is not required are likely more sponta- neous than directed, even though in Experiment 2 (and in Wahl- heim & Jacoby, 2013) participants were informed that some pairs in the second list would change, and that noticing the changes would be beneficial. Putnam et al. (2014, Experiment 2) found that such instructions had no effect on later recollection of change relative to a condition that was uninformed about changes across their “debates,” and so likely no effect on remindings, perhaps because all participants who were not informed of the relation between the two debates nonetheless noticed the repeated and changed pairs in the second debate. Even when change detection is required during study of List 2, the fact that studying the new list is the main priority, and change detection is presented as of secondary importance could tilt reminding toward spontaneous rather than directed retrieval (for a similar argument and evidence, see Kompus et al., 2011).
What controls remindings? It is tempting to argue that sponta- neous retrieval is driven by the stimulus whereas directed retrieval is driven by the task and goals of the rememberer. Characteristics of the stimuli are important for reminding as shown by the results of our logistic regression analysis. However, the task in which a person is engaged also serves an important role. Our result that remindings were more likely at slower presentation rates accords with the notion that remindings are more likely to occur when people are not narrowly focused on completing an ongoing task, that is, when attention is more diffuse (Anderson, Jacoby, Thomas, & Balota, 2011). Berntsen, Staugaard, and Sørensen (2013) re- ported results that accord with diary studies of involuntary mem- ory by showing that involuntary retrieval generally occurs during “boring tasks” with low cognitive demands.
The sensitivity of spontaneous retrieval to presentation duration and task demands may be the cause of failures to see spontaneous recovery of memories in some prior experiments. For example, Handy and Smith (2012) presented vignettes with titles and then reduced the accessibility of the critical vignettes by presenting many repetitions of filler vignettes. Later, although line drawings successfully cued deliberate recall of the titles of critical vignettes, the titles presented during a naming task failed to cue spontaneous recall. The current results show that remindings take time.
In work on proactive interference, better memory for the List 1 pair produced by multiple repetitions (Wahlheim & Jacoby, 2013), or a test of the List 1 pair prior to study of the second list (Wahlheim, 2014) has produced more remindings. Item differ- ences in memorability are also important for remindings (e.g., Putnam, Sungkhasettee, & Roediger, 2017; Wahlheim & Jacoby, 2013). Retrieval manipulations, such as reinstating a pair-specific context that appeared with the List 1 pair while studying the corresponding List 2 pair also increases the probability of remind-
ings (Kelley, Negley, & Gray, in preparation). Although not nec- essary to show facilitation effects (Aue, Criss, & Novak, 2017; Wahlheim, 2014, 2015), a relation between events can have the effect of increasing the likelihood of reminding, with backward associations between events likely to be particularly important (Sahakyan & Goodmon, 2007). There are also large individual differences in the probability of remindings that reflect different learning strategies such as the probability of looking back. Look- ing back can be brought under task control as a means of showing the importance of remindings (Jacoby et al., 2015). Later, we consider the importance of individual differences in remindings in the context of applied implications of our results.
Given that remindings oppose retroactive interference, one might ask how interference has ever been produced in experi- ments. In life, continual remindings would leave one lost in the past. Berntsen et al. (2013) demonstrated that involuntary memo- ries are reduced when cues are overloaded, just as is the case for deliberately retrieved memories, and they suggested that cue over- load keeps us from constantly being reminded. We are examining performance after a single trial whereas traditional investigations of retroactive interference and proactive interference involved extensive learning. Do remindings continue to accumulate, with remindings of remindings of remindings, as Hintzman (2004) proposed in his analysis of the role of remindings in frequency jugments? Or, instead, is there a limit to the depth of the recursive trace? With repetitions of the changed state of affairs, people may start being reminded only of the repetitions, rather than continue to add depth to original recursive reminding created when change was first noticed. In that vein, repetition effects also require remindings (Wahlheim, Maddox, & Jacoby, 2014), and the prob- ability of a reminding may be more likely for repetitions than for changed material (Benjamin & Tullis, 2010) because of greater cue overlap between the original event and the repetition, compared to the case of changed events.
Does reminding plays a role in other interference paradigms? For example, the paradigm used to study retrieval induced forget- ting pairs a single category cue with multiple exemplars. Retrieval induced forgetting is eliminated when more time is given to participants to encode category-exemplar pairs, perhaps because as in the current studies, reminding occurs more often with the longer encoding times and produces integrative encoding between exem- plars (Goodmon & Anderson, 2011). Recently, using the A–B, A–Br paradigm, Aue et al. (2017) found both proactive and ret- roactive facilitation which they attributed to reminding leading to “updating and integrating” the original trace with encoding of the new association. Reminding can be measured during study and recollection of reminding measured at test to shed light on the extent to which reminding counteracts interference.
Memory reactivation in functional MRI studies of interference potentially measures spontaneous reminding. In such studies, memory reactivation of earlier learning during related new learn- ing predicts resistance to retroactive interference (Kuhl, Shah, DuBrow, & Wagner, 2010). More recently, Koen and Rugg (2016) used multivariate similarity analysis in a retroactive interference paradigm where words were presented with different encoding tasks across lists, such as judging pleasantness, or judging ease of drawing, and at test the word was presented as a cue to recall the encoding task. Reactivation of the pattern of the first encoding task (such as the pattern specific to pleasantness judgments) during
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1361REMINDING IN RETROACTIVE EFFECTS OF MEMORY
study of a word a second time with a different encoding task predicted subsequent memory for the first encoding task. An important goal for future research is to assess the relation between conscious remindings and evidence of reactivation in functional MRI.
Talk-aloud protocols hold the potential to capture instances of spontaneous reminding in memory studies without requiring ex- plicit change detection, and we are exploring that in ongoing studies. Protocols are generally a nonreactive measure (Fox, Eric- sson, & Best, 2011) that can reveal conscious, verbalizable pro- cesses, which likely characterize remindings, and could do so without informing participants of the relation between originally learned and later learned materials.
Applied Implications
With regard to application, retroactive interference has been most extensively investigated in current research on the misinfor- mation effect in eyewitness testimony (Loftus, 2005): After view- ing a mock crime, misinformation is introduced in a subsequent narrative that then impairs memory for the details of the crime and even leads to mistaken reports of the misinformation. In the misinformation paradigm, noticing the changes in the misinforma- tion appears to be more likely with slower presentation of the misinformation, Tousignant, Hall, and Loftus (1986) found a re- duced effect of misinformation when participants were asked to read the misinformation at slower than their usual reading speed, perhaps because the slowed speed triggered more noticing of the discrepant details. In other cases of misinformation, the probability of reminding while reading the misinformation narrative may be high enough to produce overall retroactive facilitation in recall of the details contained in the original crime narrative, rather than the usual retroactive interference (Chandler, Gargano, & Holt, 2001). Recently, Putnam et al. (2017) demonstrated that when change was detected in the misinformation narratives, memory for the original crime details was enhanced.
Retroactive interference is a potential problem in education, for example, where one doesn’t want to forget prior learning in a history class of the causes of the Revolutionary War upon learning causes of the Civil War (Bower, 1974). In such cases, being reminded of the earlier learning by the new learning can serve to strengthen and so maintain the earlier learning. But the act of being reminded also brings old learning and new learning together in the current experience, so that one can mine insights from the earlier learning and apply them to the new learning, or retrospectively reassess the old learning in light of differences between the old and the new.
Reminding is also important for comprehension, both on the microlevel of noticing relations in a single text, and on the mac- rolevel of noticing relations across problems that may be separated by weeks or months. As described at the beginning of this article, Otero and Kintsch (1992) found that many students reading para- graphs that contained contradictory statements failed to notice the contradictions. Those few who did notice contradiction often re- membered both the original and the contradictory statement whereas those who did not notice the contradiction recalled only the original or the contradictory statement, or neither. The parallel to results of our experiment are obvious. On a macrolevel, research on remindings in problem solving has often found a surprising lack
of spontaneous reminding of prior problems while solving an analogous problem (e.g., Gick & Holyoak, 1983), likely due to not thinking about the subsequent problems in terms of the underlying relational structures that hold in both problems (Gentner, Loew- enstein, Thompson, & Forbus, 2009). An interesting possibility is that performance in our task might produce results that correlate with performance in a task such as that used by Otero and Kintsch or in problem solving tasks. More generally, individual differences in noticing and recollection of change as well as similarities might serve as a diagnostic device for understanding academic difficul- ties.
Interference from the distant past that limits access to the near past (proactive interference) can result in problems for adjustment, as can interference from the near past for access to the far past (retroactive interference). It is important to determine whether the two forms of interference rely on the same or different underlying mechanisms when developing treatment. Retroactive interference and reminding likely play a role in therapeutic contexts, such as cognitive– behavioral therapy where people are learning new ways of thinking and behaving. Harvey et al. (2014) notes that clients can forget psychosocial treatments when they have subsequent experiences that reinforce their maladaptive coping strategies. Anticipating such negative experiences and deliberately turning them into cues to trigger a reminding of the new self-image or coping strategy might be a fruitful strategy. In contrast, sometimes it is important to stop being reminded of prior events, as in the case of posttraumatic stress disorder, where the traumatic memory is highly likely to come to mind involuntarily because of the emo- tional intensity of the original event (Berntsen & Rubin, 2008; Staugaard & Berntsen, 2014).
References
Anderson, B. A., Jacoby, L. L., Thomas, R. C., & Balota, D. A. (2011). The effects of age and divided attention on spontaneous recognition. Memory & Cognition, 39, 725–735. http://dx.doi.org/10.3758/s13421-010- 0046-z
Anderson, M. C. (2003). Rethinking interference theory: Executive control and the mechanisms of forgetting. Journal of Memory and Language, 49, 415– 445. http://dx.doi.org/10.1016/j.jml.2003.08.006
Aue, W. R., Criss, A. H., & Novak, M. D. (2017). Evaluating mechanisms of proactive facilitation in cued recall. Journal of Memory and Lan- guage, 94, 103–118. http://dx.doi.org/10.1016/j.jml.2016.10.004
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1). http://dx.doi.org/10.18637/jss.v067.i01
Benjamin, A. S., & Ross, B. H. (2011). The causes and consequences of reminding. In A. S. Benjamin (Ed.), Successful remembering and suc- cessful forgetting: A Festschrift in honor of Robert A. Bjork (pp. 71– 88). New York, NY: Psychology Press.
Benjamin, A. S., & Tullis, J. (2010). What makes distributed practice effective? Cognitive Psychology, 61, 228 –247. http://dx.doi.org/10 .1016/j.cogpsych.2010.05.004
Berntsen, D., & Rubin, D. C. (2008). The reappearance hypothesis revis- ited: Recurrent involuntary memories after traumatic events and in everyday life. Memory & Cognition, 36, 449 – 460. http://dx.doi.org/10 .3758/MC.36.2.449
Berntsen, D., Staugaard, S. R., & Sørensen, L. M. T. (2013). Why am I remembering this now? Predicting the occurrence of involuntary (spon- taneous) episodic memories. Journal of Experimental Psychology: Gen- eral, 142, 426 – 444. http://dx.doi.org/10.1037/a0029128
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1362 NEGLEY, KELLEY, AND JACOBY
Bower, G. H. (1974). Selective facilitation and interference in retention of prose. Journal of Educational Psychology, 66, 1– 8. http://dx.doi.org/10 .1037/h0035802
Burton, R. L., Lek, I., & Caplan, J. B. (2017). Associative independence revisited: Competition between conflicting associations can be resolved or even reversed in one trial. Quarterly Journal of Experimental Psy- chology: Human Experimental Psychology, 70, 832– 857. http://dx.doi .org/10.1080/17470218.2016.1171886
Chandler, C. C., Gargano, G. J., & Holt, B. C. (2001). Witnessing poste- vents does not change memory traces, but can affect their retrieval. Applied Cognitive Psychology, 15, 3–22. http://dx.doi.org/10.1002/ 1099-0720(200101/02)15:1�3::AID-ACP669 3.0.CO;2-1
Delprato, D. J. (2005). Retroactive interference as a function of degree of interpolated study without overt retrieval practice. Psychonomic Bulletin & Review, 12, 345–349. http://dx.doi.org/10.3758/BF03196383
Fox, M. C., Ericsson, K. A., & Best, R. (2011). Do procedures for verbal reporting of thinking have to be reactive? A meta-analysis and recom- mendations for best reporting methods. Psychological Bulletin, 137, 316 –344. http://dx.doi.org/10.1037/a0021663
Gentner, D., Loewenstein, J., Thompson, L., & Forbus, K. D. (2009). Reviving inert knowledge: Analogical abstraction supports relational retrieval of past events. Cognitive Science, 33, 1343–1382. http://dx.doi .org/10.1111/j.1551-6709.2009.01070.x
Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15, 1–38. http://dx.doi.org/10.1016/ 0010-0285(83)90002-6
Goodmon, L. B., & Anderson, M. C. (2011). Semantic integration as a boundary condition on inhibitory processes in episodic retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37, 416 – 436. http://dx.doi.org/10.1037/a0021963
Halamish, V., & Bjork, R. A. (2011). When does testing enhance retention? A distribution-based interpretation of retrieval as a memory modifier. Journal of Experimental Psychology: Learning, Memory, and Cogni- tion, 37, 801– 812. http://dx.doi.org/10.1037/a0023219
Handy, J. D., & Smith, S. M. (2012). Triggering memory recovery: Effects of direct and incidental cuing. Consciousness and Cognition, 21, 1711– 1724. http://dx.doi.org/10.1016/j.concog.2012.10.003
Harvey, A. G., Lee, J., Williams, J., Hollon, S. D., Walker, M. P., Thompson, M. A., & Smith, R. (2014). Improving outcome of psycho- social treatments by enhancing memory and learning. Perspectives on Psychological Science, 9, 161–179. http://dx.doi.org/10.1177/ 1745691614521781
Hintzman, D. L. (1972). On testing the independence of associations. Psychological Review, 79, 261–264. http://dx.doi.org/10.1037/h0032684
Hintzman, D. L. (2004). Judgment of frequency versus recognition confi- dence: Repetition and recursive reminding. Memory & Cognition, 32, 336 –350. http://dx.doi.org/10.3758/BF03196863
Hintzman, D. L. (2011). Research strategy in the study of memory: Fads, fallacies, and the search for the “coordinates of truth.” Perspectives on Psychological Science, 6, 253–271. http://dx.doi.org/10.1177/ 1745691611406924
Jacoby, L. L., Kelley, C. M., & McElree, B. D. (1999). The role of cognitive control: Early selection versus late correction. In S. Chaiken & Y. Trope (Eds.), Dual process theories in social psychology (pp. 383– 400). New York, NY: Guilford Press.
Jacoby, L. L., & Wahlheim, C. N. (2013). On the importance of looking back: The role of recursive remindings in recency judgments and cued recall. Memory & Cognition, 41, 625– 637. http://dx.doi.org/10.3758/ s13421-013-0298-5
Jacoby, L. L., Wahlheim, C. N., & Kelley, C. M. (2015). Memory conse- quences of looking back to notice change: Retroactive and proactive facilitation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 41, 1282–1297. http://dx.doi.org/10.1037/xlm0000123
Jacoby, L. L., Wahlheim, C. N., & Yonelinas, A. P. (2013). The role of detection and recollection of change in list discrimination. Memory & Cognition, 41, 638 – 649. http://dx.doi.org/10.3758/s13421-013-0313-x
Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitor- ing. Psychological Bulletin, 114, 3–28. http://dx.doi.org/10.1037/0033- 2909.114.1.3
Koen, J. D., & Rugg, M. D. (2016). Memory reactivation predicts resis- tance to retroactive interference: Evidence from multivariate classifica- tion and pattern similarity analyses. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 36, 4389 – 4399. http:// dx.doi.org/10.1523/JNEUROSCI.4099-15.2016
Kompus, K., Eichele, T., Hugdahl, K., & Nyberg, L. (2011). Multimodal imaging of incidental retrieval: The low route to memory. Journal of Cognitive Neuroscience, 23, 947–960. http://dx.doi.org/10.1162/jocn .2010.21494
Kuhl, B. A., Shah, A. T., DuBrow, S., & Wagner, A. D. (2010). Resistance to forgetting associated with hippocampus-mediated reactivation during new learning. Nature Neuroscience, 13, 501–506. http://dx.doi.org/10 .1038/nn.2498
Loftus, E. F. (2005). Planting misinformation in the human mind: A 30-year investigation of the malleability of memory. Learning & Mem- ory, 12, 361–366. http://dx.doi.org/10.1101/lm.94705
Lohnas, L. J., Polyn, S. M., & Kahana, M. J. (2015). Expanding the scope of memory search: Modeling intralist and interlist effects in free recall. Psychological Review, 122, 337–363. http://dx.doi.org/10.1037/ a0039036
Martin, E. (1971). Verbal learning theory and independent retrieval phe- nomenon. Psychological Review, 78, 314 –332. http://dx.doi.org/10 .1037/h0031030
McGeoch, J. A. (1932). Forgeting and the law of disuse. Psychological Review, 39, 352–370. http://dx.doi.org/10.1037/h0069819
Melton, A. W., & Irwin, J. M. (1940). The influence of degree of inter- polated learning on retroactive inhibition and the overt transfer of specific responses. The American Journal of Psychology, S3, 173–203. http://dx.doi.org/10.2307/1417415
Mensink, G.-J., & Raaijmakers, J. G. W. (1988). A model for interference and forgetting. Psychological Review, 95, 434 – 455. http://dx.doi.org/10 .1037/0033-295X.95.4.434
Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University of South Florida word association, rhyme, and word fragment norms. Retrieved from http://www.usf.edu/FreeAssociation/
Otero, J., & Kintsch, W. (1992). Failures to detect contradictions in a text: What readers believe versus what they read. Psychological Science, 3, 229 –235. http://dx.doi.org/10.1111/j.1467-9280.1992.tb00034.x
Postman, L., & Underwood, B. J. (1973). Critical issues in interference theory. Memory & Cognition, 1, 19 – 40. http://dx.doi.org/10.3758/ BF03198064
Putnam, A. L., Sungkhasettee, V. W., & Roediger, H. L., III. (2017). When misinformation improves memory: The effects of recollecting change. Psychological Science, 28, 36 – 46. http://dx.doi.org/10.1177/ 0956797616672268
Putnam, A. L., Wahlheim, C. N., & Jacoby, L. L. (2014). Memory for flip-flopping: Detection and recollection of political contradictions. Memory & Cognition, 42, 1198 –1210. http://dx.doi.org/10.3758/ s13421-014-0419-9
R Core Team. (2013). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org/
Raaijmakers, J. G., & Shiffrin, R. M. (1981). Search of associative mem- ory. Psychological Review, 88, 93–134. http://dx.doi.org/10.1037/0033- 295X.88.2.93
Sahakyan, L., & Goodmon, L. B. (2007). The influence of directional associations on directed forgetting and interference. Journal of Experi-
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1363REMINDING IN RETROACTIVE EFFECTS OF MEMORY
mental Psychology: Learning, Memory, and Cognition, 33, 1035–1049. http://dx.doi.org/10.1037/0278-7393.33.6.1035
Schlagman, S., & Kvavilashvili, L. (2008). Involuntary autobiographical memories in and outside the laboratory: How different are they from voluntary autobiographical memories? Memory & Cognition, 36, 920 – 932. http://dx.doi.org/10.3758/MC.36.5.920
Sederberg, P. B., Gershman, S. J., Polyn, S. M., & Norman, K. A. (2011). Human memory reconsolidation can be explained using the temporal context model. Psychonomic Bulletin & Review, 18, 455– 468. http://dx .doi.org/10.3758/s13423-011-0086-9
Staugaard, S. R., & Berntsen, D. (2014). Involuntary memories of emo- tional scenes: The effects of cue discriminability and emotion over time. Journal of Experimental Psychology: General, 143, 1939 –1957. http:// dx.doi.org/10.1037/a0037185
Tousignant, J. P., Hall, D., & Loftus, E. F. (1986). Discrepancy detection and vulnerability to misleading postevent information. Memory & Cog- nition, 14, 329 –338. http://dx.doi.org/10.3758/BF03202511
Tulving, E., & Watkins, M. J. (1974). On negative transfer: Effects of testing one list on the recall of another. Journal of Verbal Learning and Verbal Behavior, 13, 181–193. http://dx.doi.org/10.1016/S0022- 5371(74)80043-5
Underwood, B. J. (1945). The effect of successive interpolations on retro- active and proactive inhibition. Psychological Monographs, 59, 1–33. http://dx.doi.org/10.1037/h0093547
Underwood, B. J., & Lund, A. M. (1981). The effect of degree of learning on within-list competition. The American Journal of Psychology, 94, 195–207. http://dx.doi.org/10.2307/1422741
Wahlheim, C. N. (2014). Proactive effects of memory in young and older adults: The role of change recollection. Memory & Cognition, 42, 950 –964. http://dx.doi.org/10.3758/s13421-014-0411-4
Wahlheim, C. N. (2015). Testing can counteract proactive interference by integrating competing information. Memory & Cognition, 43, 27–38. http://dx.doi.org/10.3758/s13421-014-0455-5
Wahlheim, C. N., & Jacoby, L. L. (2013). Remembering change: The critical role of recursive remindings in proactive effects of memory. Memory & Cognition, 41, 1–15. http://dx.doi.org/10.3758/s13421-012- 0246-9
Wahlheim, C. N., Maddox, G. B., & Jacoby, L. L. (2014). The role of reminding in the effects of spaced repetitions on cued recall: Sufficient but not necessary. Journal of Experimental Psychology: Learning, Mem- ory, and Cognition, 40, 94 –105. http://dx.doi.org/10.1037/a0034055
Wichuwut, C., & Martin, E. (1971). Independence of A–B and A–C associates in retroaction. Journal of Verbal Learning and Verbal Behav- ior, 10, 316 –321. http://dx.doi.org/10.1016/S0022-5371(71)80061-0
Winograd, E. (1968). List differentiation as a function of frequency and retention interval. Journal of Experimental Psychology, 76(2, Part 2), 1–18.
Received October 31, 2016 Revision received August 28, 2017
Accepted October 17, 2017 �
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
ti on
or on
e of
it s
al li
ed pu
bl is
he rs
. T
hi s
ar ti
cl e
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
1364 NEGLEY, KELLEY, AND JACOBY
- The Importance of Time to Think Back: The Role of Reminding in Retroactive Effects of Memory
- Experiment 1
- Method
- Participants
- Design and materials
- Procedure
- Results and Discussion
- Correct recall and intrusion errors
- Change detection and reminding
- Reminding as a source of retroactive facilitation
- Is recollection of the reminding necessary for retroactive facilitation?
- Linear mixed-effects analysis
- Experiment 2
- Method
- Participants
- Design, materials, and procedures
- Results and Discussion
- Correct recall and intrusion errors
- Recollection of the reminding
- Recall conditionalized on recollection of reminding
- General Discussion
- Implications for Theories of Interference
- Spontaneous Versus Directed Retrieval
- Applied Implications
- References