THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015 Vol. 68, No. 4, 625–634, http://dx.doi.org/10.1080/17470218.2015.1007149
Rapid communication Response cost to repeated displays—When previous distractors become targets Yingying Yang and Edward C. Merrill Department of Psychology, The University of Alabama, Tuscaloosa, AL, USA
A varied-target search task was used to evaluate the response cost of previous distractors becoming current targets in repeated visual search. We compared the relative contributions of distractor identity and location to producing response cost. During an exposure phase, half of the items were possible targets in each repeated display, and the other half were always distractors. Participants searched for a different target from the set of potential targets when the search displays were repeated. In the test phase of Experiments 1a and 1b, the roles of targets and distractors were reversed while the overall configuration was unchanged. Results indicated significant contextual costs after the switch of identities/ locations between distractors and targets. In the test phase of Experiments 2a and 2b, target identities were changed again but the target locations remained the same. Less response cost was observed in this condition relative to when both identities and locations were changed. Proximity between target and distractors in the repeated displays also influenced response cost. The mechanisms responsible for the various response cost effects and the interplay between identity, location, and proximity in the production of response cost were discussed. Keywords: Repeated visual search; Contextual cueing; Response cost; Varied-target.
It is common for people to search the same environment multiple times during their daily activities. When the environment accurately predicts where your search target will be, you can find it more easily on future searches. In an empirical demonstration, Chun and Jiang (1998) have referred to this phenomenon as contextual cueing. Search facilitation due to contextual cueing has been extensively studied in recent years. Contextual cueing effects are easily established, associated with a wide range of stimuli and conditions, attributable to both distractor identity and location, and quite robust (see Huang & Grossberg, 2010). It has also been shown that established associations between a repeated
distractor configuration and a particular target location do not readily transfer to other situations where the target location of the repeated display changes (Chun & Jiang, 1998, Experiment 6; Conci, Sun, & Müller, 2011; Kunar & Wolfe, 2011, Experiments 3–5; Zellin, Conci, van Mühlenen, & Müller, 2013). Hence, it is important to identify the specific factors that enhance or interfere with learning from repeated search. Contextual cueing effects may be accompanied by slower response times to locations occupied by distractors. Ogawa, Takeda, and Kumada (2007) found slower response times to probes presented at distractor locations of a repeated than of a new display. Similarly, Makovski and Jiang (2010)
Correspondence should be addressed to Edward C. Merrill, Department of Psychology, The University of Alabama, Box 870348, Tuscaloosa, AL 35487–0348, USA. E-mail: [email protected] © 2015 The Experimental Psychology Society
625
YANG AND MERRILL
found slower reaction times to repeated displays than new displays, when the location of the target was moved to distant previous distractor locations. This was in contrast to switching target and distractor locations that were relatively close to each other where facilitation from repeated exposure simply disappeared. They concluded that contextual cueing was driven by associative learning between the layout of the distractors and the target location. When a repeated search display was encountered, attention was first directed to the previous target location. If the target was not there, search began from that location and gradually expanded to adjacent and then more remote regions. When the new target locations were close to the learned target locations, there would still be a facilitative effect, though reduced. However, when the new target locations were relatively further away, reaction times exceeded those for the new displays and resulted in a contextual cost. To date, because most of the available research on repeated visual search has focused on facilitation effects, very little is known about the specific properties of response cost in repeated visual search. In the research reported here, we focused on the type of stimulus information that results in response cost during repeated visual search. More specifically, we were interested in whether contextual cost may be associated with both individual distractor identities and distractor locations. Research has shown that facilitation can be associated with both properties (e.g., Endo & Takeda, 2004; Jiang & Song, 2005). Hence, it is reasonable to expect that both may be associated with response cost. A process that inhibits individual distractor identities and locations through repeated exposure would probably involve a relatively extensive memory of the repeated displays that requires keeping track of specific items/locations that had been targets over time relative to those that had never been targets. To address this issue, we developed a new variable-target repeated search paradigm. Our procedure required participants to search for different targets in varied locations in each search display across repeated exposures. We expected this procedure to provide a greater opportunity for
626
processing individual distractor identities/locations as learning progressed. Each display contained 16 letters. During the repeated exposures, participants searched for eight of the letters (one designated letter per trial) with the other eight letters always being distractors. Following the exposure phase, the previous distractor letters became the new targets in the test phase. Response times to locating these new targets in the changed repeated displays were compared to locating these letters in new displays. Longer response times in the changed repeated displays would reflect response cost. There were several important differences between previous studies using single targets and homogeneous distractors (e.g., Makovski & Jiang, 2010) and our studies. First, the targets in our test phase changed identity as well as location. Second, we required participants to view and remember a different target identity at the start of each trial. This probably would prevent target response facilitation in the form of contextual cueing effects from being observed. Third, the overall spatial layout of our displays was always a square, and only internal relationships between letters could vary, eliminating the predictability of overall spatial configuration. Fourth, targets and distractors were distributed throughout the repeated displays, and attention should be evenly spread across the entire display rather than having a spotlight focused around one target. Hence, our procedure probably contributed to forming target–distractor identity associations as well as target–distractor location associations. Fifth, in contrast to most studies of repeated visual search, we had participants locate the target rather than respond on the basis of some other facets of the target (facing left/right). Nevertheless, each display was identified with at least half, if not all, of the possible responses. Therefore, it was not possible for participants to form meaningful response–display associations; instead, repeated– new differences would necessarily be attributed to display content. We manipulated one additional variable. Experiments 1a and 1b differed with respect to the proximity of potential targets and distractors in the learning phase. In Experiment 1a, targets
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
and distractors shared the same quadrants of the repeated displays. In Experiment 1b, targets and distractors were always in different quadrants. Makovski and Jiang (2010) reported greater response cost when targets were moved to distant locations. Contrasting Experiment 1a with Experiment 1b would reveal whether this distance effect could also be observed when both location and identity aspects of the target changed in our variable-target paradigm. This manipulation would provide some insight into the interrelations between distractor identity and location in repeated visual search.
EXPERIMENT 1A Method Participants Participants in all experiments were Introductory Psychology students (17–22 years old) who earned course credit for participating. In Experiment 1a, 32 students participated. Materials Experimental materials were displays of 16 English letters presented on a computer monitor. As shown in Figure 1, four letters were grouped in each quadrant of the displays. Six configurations were constructed for use as “repeated displays” in which the letters remained in the same locations throughout the exposure phase. Eight letters in each repeated display, two in each quadrant, were selected as
possible targets before the experiment. One of these eight was designated as the target on each exposure trial. The other eight letters were never targets during the exposure phase. Hence, for each presentation of each repeated display, the arrangement of individual letters was identical, the target was selected from the pool of the eight potential targets, and the remaining eight letters were always distractors. Different letters were designated as targets for the six different repeated displays, with each letter appearing as a target the same number of times across presentations during the exposure phase. The eight distractors and targets switched roles in each repeated display for the test phase (see Figure 2). These were named different target and location (DTL) displays because the new target was both a different letter and in a different location from any of the targets in the exposure phase. Forty-eight new displays were also constructed for the test phase with the same 16 letters presented randomly in the four quadrants. Procedure There was an exposure phase and a test phase. The exposure phase was divided into five blocks. Each of the six repeated displays appeared eight times per block, and each possible target letter was associated with its repeated display once per block. The test phase (Block 6) consisted of 96 trials: the 48 DTL displays and 48 new displays. Trials were randomized within blocks. Trials started with a fixation cross for 750 ms, followed by a letter identifying the target for the upcoming trial. The letter remained visible until
Figure 1. Examples of three repeated configurations. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
627
YANG AND MERRILL
participants chose an incorrect quadrant or failed to respond within 10 s, a beep was heard, and an error was recorded. Response times were recorded to the nearest millisecond.
Results Figure 2. Examples of one repeated display in Experiment 1. In this example, both figures have the same repeated configurations. In Figure A the circled letters represent the potential targets in the exposure phase. They were evenly distributed throughout the display. In Figure B the circled letters represent the potential targets in the test phase.
Errors were rare. The data from two participants who committed over 5% errors were excluded from the final analysis. The error rate for the final sample was 3.5% and was not analysed. Mean reaction times (RTs) of correct responses were calculated for the repeated displays in Blocks 1–5 and the DTL and new displays in Block 6 (see Figure 4). RTs in exposure blocks were analysed via a repeated measures analysis of variance (ANOVA). There was a main effect of block, F(2.24, 64.81) = 10.52 (Greenhouse–Geisser), p , .001, h2p = .266, with RTs decreasing over blocks. RTs of DTL and new displays in the test phase were also analysed using a repeated measures ANOVA. Participants responded 48 ms slower to the DTL displays (1628 ms) than the new displays (1580 ms), F(1, 29) = 5.20, p = .030, h2p = .152, indicating that RTs were slower when previous distractor letters/ locations became targets.
Figure 3. The basic procedure of Experiments 1 and 2. Participants responded by identifying the quadrant in which the target appeared using the keyboard (R = upper left, I = upper right, F = lower left, and J = lower right).
the participant pressed the spacebar. The search display was presented immediately. Participants pressed a designated key indicating the quadrant where the target was located (see Figure 3). If
628
Figure 4. Response times (ms) of repeated displays in the exposure and test phases of Experiment 1a. B1–B5 refer to repeated displays in Blocks 1–5 of the exposure phase. DTL (different target and location) and “new” refer to the test phase.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
EXPERIMENT 1B Method Participants Thirty-two different students participated.
Materials and procedure Repeated and new displays were identical to those in Experiment 1a except that the eight target locations were concentrated in two quadrants in each repeated display during the exposure phase (see Figure 5). In the test phase, targets in each DTL display were in the other two quadrants.
Results Two participants made over 5% of errors, and their data were not included in the analysis. The error rate for the final sample was 2.4%. Mean RTs were calculated as in Experiment 1a (see Figure 6). Analysis of RTs of repeated displays in the exposure phase revealed a main effect of block, F(3.38, 98.00) = 19.97, p , .001, h2p = .397, with RTs decreasing over blocks. In the test phase, participants responded significantly slower by 78 ms to the DTL (1668 ms) than to the new displays (1590 ms), F(1, 29) = 9.31, p = .005, h2p = .243. Hence, a response cost was also observed in Experiment 1b.
Figure 6. Response times (ms) of repeated displays in the exposure and test phases of Experiment 1b. B1–B5 refers to repeated displays in Blocks 1–5 of the exposure phase. DTL (different target and location) and “new” refer to the test phase.
Cross-experiment analyses We conducted an Experiment (1a vs. 1b) × Repetition (DTL vs. new) ANOVA on RTs in the test phase to determine whether grouping distractors away from targets impacted response cost. The main effect of repetition was significant, F(1, 58) = 14.48, p , .001, h2p = .200. However, neither the main effect of experiment nor the Experiment × Repetition interaction was significant. Hence, response cost was statistically similar across experiments (48 vs. 78 ms).
Discussion
Figure 5. Examples of one repeated display in Experiment 1b. In this example, both figures have the exact same repeated configurations. In Figure A the circled letters represent the potential targets in the exposure phase. They were clustered into two quadrants. In Figure B the circled letters represent the potential targets in the test phase. Targets and distractors switched in the test phase.
In Experiments 1a and 1b, we found that switching target and distractor identities/locations resulted in a significant contextual cost. This was true whether targets were switched to near or far distractor locations. For at least the near locations, it seems likely that cost was associated with individual distractor identities and locations because the original targets and distractors were intermixed within quadrants. Despite not being significant, there was a larger inhibition effect in Experiment 1b than in Experiment 1a. Hence, when the original targets and distractors were separated by different quadrants, it is possible that a portion of the cost could
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
629
YANG AND MERRILL
be due to participants inhibiting larger areas (e.g., entire quadrants) of the repeated displays that never contained targets. This may also be attributable to the fact that participants were actually making a different response to the targets in DTL condition of Experiment 1b because changed targets were in different quadrants from those for original targets. Importantly, inhibition was observed in both studies indicating that response cost can be associated with specific distractor identity/locations in addition to general locations. In Experiments 2a and 2b, we evaluated whether the observed response cost was largely identity or location driven.
EXPERIMENT 2 The response cost observed in Experiment 1 could be attributed to two different mechanisms. First, it could be that participants inhibited distractors on the basis of their identity. Second, it could be that participants inhibited the distractor locations rather than or in addition to their identities. To evaluate these possibilities, we conducted two additional experiments. In Experiments 2a and 2b, we again included DTL displays to provide a baseline for comparison and replicate the novel results of Experiment 1. In addition, we included displays where we presented changed target identities in an old target location (different target only, DTO, displays). If the identity–context association plays an important role in the response cost effect, then significantly slower RTs relative to new displays should be observed in all four change conditions (DTL and DTO of Experiments 2a and 2b) because target identity changed in all conditions. If the location–context association is also important in producing response cost, then reduced contextual cost should be observed in the DTO conditions because target location was unchanged.
Materials and procedure Experiment 2a included two target change conditions in the test phase. There was a DTL condition that was identical to the DTL condition of Experiment 1a; a previous distractor became the target in its previous location. In addition, there was a DTO condition in which a previous distractor was moved to the same location as a previous target and became the new target (see Figure 7). One additional constraint was that the moved distractor and the target location that it replaced were adjacent to each other in the repeated displays. Each target location of the original display was replaced by only one distractor. For each of the six repeated configurations, eight DTO displays were created (see Figure 7). In the test phase, there were 48 DTL displays, 48 DTO, displays and 48 new displays. All other aspects of the procedure were identical to those in Experiment 1a.
EXPERIMENT 2B Method Participants Twenty-seven additional students participated. Materials and procedure Experiment 2b was identical to Experiment 1b except for the addition of the DTO condition. In the DTO condition, the previous distractor was moved to the same location as a previous target in a different quadrant and became the new target (see Figure 8). As in Experiment 2a, there were three types of displays in the test phase of Experiment 2b: 48 DTL, 48 DTO, and 48 new displays.
Results EXPERIMENT 2A Method Participants Twenty-nine students participated.
630
Experiments 2a and 2b were analysed together. Again errors were rare. Two participants’ data from Experiment 2a were not included in the analysis: One failed to complete the task, and one committed errors exceeding 20%. Error rates for
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
Figure 7. Examples of different target only (DTO) display in Experiment 2a and its construction. Figure A is the repeated display in the exposure phase with circles highlighting potential targets. Figure B is one example of the same repeated configuration where target was F. Figure C depicts the creation of the DTO display where the previous target switched identity with a previous distractor in the same quadrant. Figure D is the newly created DTO display where the target was circled as letter A.
Figure 8. Examples of different target only (DTO) display in Experiment 2b and its construction. Figure A is the repeated display in the exposure phase with circles highlighting potential targets. Figure B is one example of the same repeated configuration where target was F. Figure C depicts the creation of the DTO displays where the previous target switched identity with a previous distractor in a different quadrant. Figure D is the newly created DTO display where the target was circled as letter Y. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
631
YANG AND MERRILL
the final sample was again under 5% for both experiments and were not analysed. Means for each block and condition are reported in Figures 9 and 10. The exposure phases were analysed using an Experiment (2a vs. 2B) × Blocks (1–5) mixed ANOVA. The analysis revealed a F(3.20, 166.44) = 8.53, p , .001, h2p = .141, with
Figure 9. Response times (ms) of displays in the exposure and test phases of Experiment 2a. B1–B5 refer to repeated displays in Blocks 1–5 in the exposure phase. DTL (different target and location), DTO (different target only), and “new” refer to the test phase.
RTs decreasing over exposure blocks. Neither the main effect of experiment nor the Experiment × Block interaction was significant. We then conducted preliminary analyses using paired-sample t-tests to determine whether response cost was observed in any of the four conditions (by comparing each to the corresponding new condition of their respective experiments). These analyses (see Table 1) revealed a significant response cost in three of the four repeated conditions: the DTL condition of Experiment 2a, t(26) = 2.361, p = .026, the DTL condition of Experiment 2b, t(26) = 3.816, p = .001, and the DTO condition of Experiment 2b, t(26) = 3.282, p = .003. However, response cost in the DTO condition of Experiment 2a was not significant. To address our primary hypotheses, we converted the RTs to measures of response cost by subtracting RTs for the new displays from those of the DTL and DTO conditions of each experiment separately. These scores were compared using an Experiment (2a vs. 2b) × Location (DTO vs. DTL) mixed ANOVA. The analysis revealed a marginally significant main effect of experiment, F(1, 52) = 3.59, p = .064, h2p = .065, and a significant main effect of location, F(1, 52) = 4.13, p = .047, h2p = .074. The interaction was not significant. Taken together, the two main effects indicated that there was greater response cost when the distractor turned target was more distant from previous target locations (in another quadrant) than when it was close to previous target locations in the learning phase. In addition, there was greater response cost associated with distractors that remained in a distractor location than with those that moved to a previous target location in the test phase.
GENERAL DISCUSSION
Figure 10. Response times (ms) of displays in the exposure and test phases of Experiment 2b. B1–B5 refer to repeated displays in Blocks 1–5 in the exposure phase. DTL (different target and location), DTO (different target only), and “new” refer to the test phase.
632
We investigated the response cost associated with previous distractors becoming targets in a variedtarget repeated search paradigm. Across four conditions, we disrupted the target–identity context association by switching target with distractor identity/location combinations or target with distractor
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
Table 1. Mean reaction times (in ms) of displays in test phase of Experiments 2a and 2b Displays Experiment 2a 2b
New difference
New
DTL
DTO
DTL
DTO
1411 (175) 1439 (208)
1452 (158) 1535 (218)
1430 (172) 1496 (190)
41* (p = .026) 96** (p = .001)
19 (p = .894) 57** (p = .003)
Note: DTL = different target and location; DTO = different target only. Standard deviations and significance in parentheses. p* , .05. p** , .01.
identity only. In most conditions, participants responded significantly more slowly to repeated displays than to new displays when distractors in the repeated displays were designated as new targets in the test phase. More specifically, in Experiments 1a and 1b, the target in the test phase changed identity and location and resulted in significant response cost. In Experiments 2a and 2b, we compared targets that changed in both features to those that changed in identity but not location. Significant response cost was observed in both combinations. However, there was significantly less cost when the new target identity appeared in an old target location. In fact, in the DTO condition of Experiment 2a (same quadrant) we did not observe significant response cost. This result is addressed in detail below. Nevertheless, it appears that distractor identity and location generally contribute to the observation of response cost. From repeated exposure, participants appear to learn that some identity/ location combinations are never targets. This knowledge is expressed as slower responding to those identities/locations once they became the new targets, because the contextual or statistical past (Zellin et al., 2013) of the previous-distractor identity/location is not maintained in the test phase. Even so, the magnitude of that cost is susceptible to the influence of at least two variables: the proximity of distractors to a target during repeated search and whether the new target appeared in the same location as the old target. Distractors that were closer to an original target position generated less response cost than did distractors that were further from the target.
However, the mechanisms may be different from those of Makovski and Jiang (2010), who attributed response cost to search time from the original target position. In our study, targets were distributed throughout the display, and there was no single target position to begin a search for the new targets. Because distractors were interspersed with and near the targets in the learning phase in Experiments 1a and 2a, distractor identities were learned and associated with a specific context, although probably to a lesser degree than the target. It is reasonable to suspect that distractors near a target received some positive spillover during search (Brady & Chun, 2007). This resulted in a reduction in response cost relative to Experiments 1b and 2b in which distractors were probably learned to a lesser degree because they were more distant from any target and congregated in different regions of the displays. We also observed that changes in identity alone resulted in less cost than changes in both identity and location. It appears that having the new target appear in a previous target location may counterbalance some of the inhibition accrued for previous distractor identities. Hence, the lack of significant contextual cost in DTO condition in Experiment 2a may be due to a combination of opposing effects. There is a general response cost that we attribute to a disruption in learned identity–context associations. Cost was reduced when the distractor turned target appeared in a previous target location. As mentioned above, this cost was further reduced because of proximity between targets and distractors during repeated search. Note, however, that significant response cost
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
633
YANG AND MERRILL
was still observed when distractors were moved to previous target locations that were more distant, indicating that cost can be produced by changes in target identity alone. Thus, location information and identity information can operate independently in determining the magnitude of response cost. Taken together, our results are consistent with general models of visual search that suggest that nontarget objects can be explicitly or implicitly inhibited based on features, similarities, or spatial locations (see Crawford, Hill, & Higham, 2005; Treisman & Gelade, 1980; Wolfe, 1994). By “marking off” distractors, the capacity-limited attention system can be more efficiently deployed to possible target locations and refrain from returning to the previously visited nontarget objects (Horowitz & Wolfe, 2003). However, learning that a particular identity/location was never a target may incur a response cost when this object becomes a target in the future. Our studies indicate a complex relationship between identity inhibition, location inhibition, and vicinity of the distractors to the previous targets in producing response cost. Robust identity-based response cost was found whether location inhibition was absent or not and whether previous distractors were close to the previous target or not, but just not when both of them were operating. Original manuscript received 19 March 2014 Accepted revision received 23 December 2014 First published online 18 February 2015
REFERENCES Brady, T. F., & Chun, M. M. (2007). Spatial constraints on learning in visual search: Modeling contextual cuing. Journal Of Experimental Psychology: Human Perception And Performance, 33(4), 798–815. doi:10. 1037/0096-1523.33.4.798 Chun, M. M., & Jiang, Y. (1998). Contextual cueing: Implicit learning and memory of visual context
634
guides spatial attention. Cognitive Psychology, 36, 28–71. Conci, M., Sun, L., & Müller, H. J. (2011). Contextual remapping in visual search after predictable targetlocation changes. Psychological Research, 75(4), 279– 289. doi:10.1007/s00426-010-0306-3 Crawford, T. J., Hill, S., & Higham, S. (2005). The inhibitory effect of a recent distracter. Vision Research, 45(27), 3365–3378. doi:10.1016/j.visres. 2005.07.024 Endo, N., & Takeda, Y. (2004). Selective learning of spatial configuration and object identity in visual search. Perception & Psychophysics, 68(2), 293–302. Horowitz, T. S., & Wolfe, J. M. (2003). Memory for rejected distractors in visual search? Visual Cognition, 10(3), 257–298. doi:10.1080/ 13506280143000005 Huang, T., & Grossberg, S. (2010). Cortical dynamics of contextually cued attentive visual learning and search: Spatial and object evidence accumulation. Psychological Review, 117(4), 1080–1112. doi:10.1037/a0020664 Jiang, Y., & Song, J. (2005). Hyperspecificity in Visual Implicit Learning: Learning of Spatial Layout Is Contingent on Item Identity. Journal of Experimental Psychology: Human Perception And Performance, 31(6), 1439–1448. doi:10.1037/00961523.31.6.1439. Kunar, M. A., & Wolfe, J. M. (2011). Target absent trials in configural contextual cuing. Attention, Perception, & Psychophysics, 73(7), 2077–2091. Makovski, T., & Jiang, Y. V. (2010). Contextual cost: When a visual-search target is not where it should be. The Quarterly Journal of Experimental Psychology, 63(2), 216–225. doi:10.1080/17470210903281590 Ogawa, H., Takeda, Y., & Kumada, T. (2007). Probing attentional modulation of contextual cueing. Visual Cognition, 15, 276–289. Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention. Cognitive Psychology, 12 (1), 97–136. Wolfe, J. M. (1994). Guided search 2.0: A revised model of visual search. Psychonomic Bulletin & Review, 1(2), 202–238. doi:10.3758/BF03200774 Zellin, M., Conci, M., van Mühlenen, A., & Müller, H. J. (2013). Here today, gone tomorrow: Memory adaptation in contextual cueing. PLoS One, 8(3), e59466. doi:10.1371/journal.pone.0059466
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
Copyright of Quarterly Journal of Experimental Psychology is the property of Psychology Press (UK) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
Rapid communication Response cost to repeated displays—When previous distractors become targets Yingying Yang and Edward C. Merrill Department of Psychology, The University of Alabama, Tuscaloosa, AL, USA
A varied-target search task was used to evaluate the response cost of previous distractors becoming current targets in repeated visual search. We compared the relative contributions of distractor identity and location to producing response cost. During an exposure phase, half of the items were possible targets in each repeated display, and the other half were always distractors. Participants searched for a different target from the set of potential targets when the search displays were repeated. In the test phase of Experiments 1a and 1b, the roles of targets and distractors were reversed while the overall configuration was unchanged. Results indicated significant contextual costs after the switch of identities/ locations between distractors and targets. In the test phase of Experiments 2a and 2b, target identities were changed again but the target locations remained the same. Less response cost was observed in this condition relative to when both identities and locations were changed. Proximity between target and distractors in the repeated displays also influenced response cost. The mechanisms responsible for the various response cost effects and the interplay between identity, location, and proximity in the production of response cost were discussed. Keywords: Repeated visual search; Contextual cueing; Response cost; Varied-target.
It is common for people to search the same environment multiple times during their daily activities. When the environment accurately predicts where your search target will be, you can find it more easily on future searches. In an empirical demonstration, Chun and Jiang (1998) have referred to this phenomenon as contextual cueing. Search facilitation due to contextual cueing has been extensively studied in recent years. Contextual cueing effects are easily established, associated with a wide range of stimuli and conditions, attributable to both distractor identity and location, and quite robust (see Huang & Grossberg, 2010). It has also been shown that established associations between a repeated
distractor configuration and a particular target location do not readily transfer to other situations where the target location of the repeated display changes (Chun & Jiang, 1998, Experiment 6; Conci, Sun, & Müller, 2011; Kunar & Wolfe, 2011, Experiments 3–5; Zellin, Conci, van Mühlenen, & Müller, 2013). Hence, it is important to identify the specific factors that enhance or interfere with learning from repeated search. Contextual cueing effects may be accompanied by slower response times to locations occupied by distractors. Ogawa, Takeda, and Kumada (2007) found slower response times to probes presented at distractor locations of a repeated than of a new display. Similarly, Makovski and Jiang (2010)
Correspondence should be addressed to Edward C. Merrill, Department of Psychology, The University of Alabama, Box 870348, Tuscaloosa, AL 35487–0348, USA. E-mail: [email protected] © 2015 The Experimental Psychology Society
625
YANG AND MERRILL
found slower reaction times to repeated displays than new displays, when the location of the target was moved to distant previous distractor locations. This was in contrast to switching target and distractor locations that were relatively close to each other where facilitation from repeated exposure simply disappeared. They concluded that contextual cueing was driven by associative learning between the layout of the distractors and the target location. When a repeated search display was encountered, attention was first directed to the previous target location. If the target was not there, search began from that location and gradually expanded to adjacent and then more remote regions. When the new target locations were close to the learned target locations, there would still be a facilitative effect, though reduced. However, when the new target locations were relatively further away, reaction times exceeded those for the new displays and resulted in a contextual cost. To date, because most of the available research on repeated visual search has focused on facilitation effects, very little is known about the specific properties of response cost in repeated visual search. In the research reported here, we focused on the type of stimulus information that results in response cost during repeated visual search. More specifically, we were interested in whether contextual cost may be associated with both individual distractor identities and distractor locations. Research has shown that facilitation can be associated with both properties (e.g., Endo & Takeda, 2004; Jiang & Song, 2005). Hence, it is reasonable to expect that both may be associated with response cost. A process that inhibits individual distractor identities and locations through repeated exposure would probably involve a relatively extensive memory of the repeated displays that requires keeping track of specific items/locations that had been targets over time relative to those that had never been targets. To address this issue, we developed a new variable-target repeated search paradigm. Our procedure required participants to search for different targets in varied locations in each search display across repeated exposures. We expected this procedure to provide a greater opportunity for
626
processing individual distractor identities/locations as learning progressed. Each display contained 16 letters. During the repeated exposures, participants searched for eight of the letters (one designated letter per trial) with the other eight letters always being distractors. Following the exposure phase, the previous distractor letters became the new targets in the test phase. Response times to locating these new targets in the changed repeated displays were compared to locating these letters in new displays. Longer response times in the changed repeated displays would reflect response cost. There were several important differences between previous studies using single targets and homogeneous distractors (e.g., Makovski & Jiang, 2010) and our studies. First, the targets in our test phase changed identity as well as location. Second, we required participants to view and remember a different target identity at the start of each trial. This probably would prevent target response facilitation in the form of contextual cueing effects from being observed. Third, the overall spatial layout of our displays was always a square, and only internal relationships between letters could vary, eliminating the predictability of overall spatial configuration. Fourth, targets and distractors were distributed throughout the repeated displays, and attention should be evenly spread across the entire display rather than having a spotlight focused around one target. Hence, our procedure probably contributed to forming target–distractor identity associations as well as target–distractor location associations. Fifth, in contrast to most studies of repeated visual search, we had participants locate the target rather than respond on the basis of some other facets of the target (facing left/right). Nevertheless, each display was identified with at least half, if not all, of the possible responses. Therefore, it was not possible for participants to form meaningful response–display associations; instead, repeated– new differences would necessarily be attributed to display content. We manipulated one additional variable. Experiments 1a and 1b differed with respect to the proximity of potential targets and distractors in the learning phase. In Experiment 1a, targets
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
and distractors shared the same quadrants of the repeated displays. In Experiment 1b, targets and distractors were always in different quadrants. Makovski and Jiang (2010) reported greater response cost when targets were moved to distant locations. Contrasting Experiment 1a with Experiment 1b would reveal whether this distance effect could also be observed when both location and identity aspects of the target changed in our variable-target paradigm. This manipulation would provide some insight into the interrelations between distractor identity and location in repeated visual search.
EXPERIMENT 1A Method Participants Participants in all experiments were Introductory Psychology students (17–22 years old) who earned course credit for participating. In Experiment 1a, 32 students participated. Materials Experimental materials were displays of 16 English letters presented on a computer monitor. As shown in Figure 1, four letters were grouped in each quadrant of the displays. Six configurations were constructed for use as “repeated displays” in which the letters remained in the same locations throughout the exposure phase. Eight letters in each repeated display, two in each quadrant, were selected as
possible targets before the experiment. One of these eight was designated as the target on each exposure trial. The other eight letters were never targets during the exposure phase. Hence, for each presentation of each repeated display, the arrangement of individual letters was identical, the target was selected from the pool of the eight potential targets, and the remaining eight letters were always distractors. Different letters were designated as targets for the six different repeated displays, with each letter appearing as a target the same number of times across presentations during the exposure phase. The eight distractors and targets switched roles in each repeated display for the test phase (see Figure 2). These were named different target and location (DTL) displays because the new target was both a different letter and in a different location from any of the targets in the exposure phase. Forty-eight new displays were also constructed for the test phase with the same 16 letters presented randomly in the four quadrants. Procedure There was an exposure phase and a test phase. The exposure phase was divided into five blocks. Each of the six repeated displays appeared eight times per block, and each possible target letter was associated with its repeated display once per block. The test phase (Block 6) consisted of 96 trials: the 48 DTL displays and 48 new displays. Trials were randomized within blocks. Trials started with a fixation cross for 750 ms, followed by a letter identifying the target for the upcoming trial. The letter remained visible until
Figure 1. Examples of three repeated configurations. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
627
YANG AND MERRILL
participants chose an incorrect quadrant or failed to respond within 10 s, a beep was heard, and an error was recorded. Response times were recorded to the nearest millisecond.
Results Figure 2. Examples of one repeated display in Experiment 1. In this example, both figures have the same repeated configurations. In Figure A the circled letters represent the potential targets in the exposure phase. They were evenly distributed throughout the display. In Figure B the circled letters represent the potential targets in the test phase.
Errors were rare. The data from two participants who committed over 5% errors were excluded from the final analysis. The error rate for the final sample was 3.5% and was not analysed. Mean reaction times (RTs) of correct responses were calculated for the repeated displays in Blocks 1–5 and the DTL and new displays in Block 6 (see Figure 4). RTs in exposure blocks were analysed via a repeated measures analysis of variance (ANOVA). There was a main effect of block, F(2.24, 64.81) = 10.52 (Greenhouse–Geisser), p , .001, h2p = .266, with RTs decreasing over blocks. RTs of DTL and new displays in the test phase were also analysed using a repeated measures ANOVA. Participants responded 48 ms slower to the DTL displays (1628 ms) than the new displays (1580 ms), F(1, 29) = 5.20, p = .030, h2p = .152, indicating that RTs were slower when previous distractor letters/ locations became targets.
Figure 3. The basic procedure of Experiments 1 and 2. Participants responded by identifying the quadrant in which the target appeared using the keyboard (R = upper left, I = upper right, F = lower left, and J = lower right).
the participant pressed the spacebar. The search display was presented immediately. Participants pressed a designated key indicating the quadrant where the target was located (see Figure 3). If
628
Figure 4. Response times (ms) of repeated displays in the exposure and test phases of Experiment 1a. B1–B5 refer to repeated displays in Blocks 1–5 of the exposure phase. DTL (different target and location) and “new” refer to the test phase.
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
EXPERIMENT 1B Method Participants Thirty-two different students participated.
Materials and procedure Repeated and new displays were identical to those in Experiment 1a except that the eight target locations were concentrated in two quadrants in each repeated display during the exposure phase (see Figure 5). In the test phase, targets in each DTL display were in the other two quadrants.
Results Two participants made over 5% of errors, and their data were not included in the analysis. The error rate for the final sample was 2.4%. Mean RTs were calculated as in Experiment 1a (see Figure 6). Analysis of RTs of repeated displays in the exposure phase revealed a main effect of block, F(3.38, 98.00) = 19.97, p , .001, h2p = .397, with RTs decreasing over blocks. In the test phase, participants responded significantly slower by 78 ms to the DTL (1668 ms) than to the new displays (1590 ms), F(1, 29) = 9.31, p = .005, h2p = .243. Hence, a response cost was also observed in Experiment 1b.
Figure 6. Response times (ms) of repeated displays in the exposure and test phases of Experiment 1b. B1–B5 refers to repeated displays in Blocks 1–5 of the exposure phase. DTL (different target and location) and “new” refer to the test phase.
Cross-experiment analyses We conducted an Experiment (1a vs. 1b) × Repetition (DTL vs. new) ANOVA on RTs in the test phase to determine whether grouping distractors away from targets impacted response cost. The main effect of repetition was significant, F(1, 58) = 14.48, p , .001, h2p = .200. However, neither the main effect of experiment nor the Experiment × Repetition interaction was significant. Hence, response cost was statistically similar across experiments (48 vs. 78 ms).
Discussion
Figure 5. Examples of one repeated display in Experiment 1b. In this example, both figures have the exact same repeated configurations. In Figure A the circled letters represent the potential targets in the exposure phase. They were clustered into two quadrants. In Figure B the circled letters represent the potential targets in the test phase. Targets and distractors switched in the test phase.
In Experiments 1a and 1b, we found that switching target and distractor identities/locations resulted in a significant contextual cost. This was true whether targets were switched to near or far distractor locations. For at least the near locations, it seems likely that cost was associated with individual distractor identities and locations because the original targets and distractors were intermixed within quadrants. Despite not being significant, there was a larger inhibition effect in Experiment 1b than in Experiment 1a. Hence, when the original targets and distractors were separated by different quadrants, it is possible that a portion of the cost could
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
629
YANG AND MERRILL
be due to participants inhibiting larger areas (e.g., entire quadrants) of the repeated displays that never contained targets. This may also be attributable to the fact that participants were actually making a different response to the targets in DTL condition of Experiment 1b because changed targets were in different quadrants from those for original targets. Importantly, inhibition was observed in both studies indicating that response cost can be associated with specific distractor identity/locations in addition to general locations. In Experiments 2a and 2b, we evaluated whether the observed response cost was largely identity or location driven.
EXPERIMENT 2 The response cost observed in Experiment 1 could be attributed to two different mechanisms. First, it could be that participants inhibited distractors on the basis of their identity. Second, it could be that participants inhibited the distractor locations rather than or in addition to their identities. To evaluate these possibilities, we conducted two additional experiments. In Experiments 2a and 2b, we again included DTL displays to provide a baseline for comparison and replicate the novel results of Experiment 1. In addition, we included displays where we presented changed target identities in an old target location (different target only, DTO, displays). If the identity–context association plays an important role in the response cost effect, then significantly slower RTs relative to new displays should be observed in all four change conditions (DTL and DTO of Experiments 2a and 2b) because target identity changed in all conditions. If the location–context association is also important in producing response cost, then reduced contextual cost should be observed in the DTO conditions because target location was unchanged.
Materials and procedure Experiment 2a included two target change conditions in the test phase. There was a DTL condition that was identical to the DTL condition of Experiment 1a; a previous distractor became the target in its previous location. In addition, there was a DTO condition in which a previous distractor was moved to the same location as a previous target and became the new target (see Figure 7). One additional constraint was that the moved distractor and the target location that it replaced were adjacent to each other in the repeated displays. Each target location of the original display was replaced by only one distractor. For each of the six repeated configurations, eight DTO displays were created (see Figure 7). In the test phase, there were 48 DTL displays, 48 DTO, displays and 48 new displays. All other aspects of the procedure were identical to those in Experiment 1a.
EXPERIMENT 2B Method Participants Twenty-seven additional students participated. Materials and procedure Experiment 2b was identical to Experiment 1b except for the addition of the DTO condition. In the DTO condition, the previous distractor was moved to the same location as a previous target in a different quadrant and became the new target (see Figure 8). As in Experiment 2a, there were three types of displays in the test phase of Experiment 2b: 48 DTL, 48 DTO, and 48 new displays.
Results EXPERIMENT 2A Method Participants Twenty-nine students participated.
630
Experiments 2a and 2b were analysed together. Again errors were rare. Two participants’ data from Experiment 2a were not included in the analysis: One failed to complete the task, and one committed errors exceeding 20%. Error rates for
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
Figure 7. Examples of different target only (DTO) display in Experiment 2a and its construction. Figure A is the repeated display in the exposure phase with circles highlighting potential targets. Figure B is one example of the same repeated configuration where target was F. Figure C depicts the creation of the DTO display where the previous target switched identity with a previous distractor in the same quadrant. Figure D is the newly created DTO display where the target was circled as letter A.
Figure 8. Examples of different target only (DTO) display in Experiment 2b and its construction. Figure A is the repeated display in the exposure phase with circles highlighting potential targets. Figure B is one example of the same repeated configuration where target was F. Figure C depicts the creation of the DTO displays where the previous target switched identity with a previous distractor in a different quadrant. Figure D is the newly created DTO display where the target was circled as letter Y. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
631
YANG AND MERRILL
the final sample was again under 5% for both experiments and were not analysed. Means for each block and condition are reported in Figures 9 and 10. The exposure phases were analysed using an Experiment (2a vs. 2B) × Blocks (1–5) mixed ANOVA. The analysis revealed a F(3.20, 166.44) = 8.53, p , .001, h2p = .141, with
Figure 9. Response times (ms) of displays in the exposure and test phases of Experiment 2a. B1–B5 refer to repeated displays in Blocks 1–5 in the exposure phase. DTL (different target and location), DTO (different target only), and “new” refer to the test phase.
RTs decreasing over exposure blocks. Neither the main effect of experiment nor the Experiment × Block interaction was significant. We then conducted preliminary analyses using paired-sample t-tests to determine whether response cost was observed in any of the four conditions (by comparing each to the corresponding new condition of their respective experiments). These analyses (see Table 1) revealed a significant response cost in three of the four repeated conditions: the DTL condition of Experiment 2a, t(26) = 2.361, p = .026, the DTL condition of Experiment 2b, t(26) = 3.816, p = .001, and the DTO condition of Experiment 2b, t(26) = 3.282, p = .003. However, response cost in the DTO condition of Experiment 2a was not significant. To address our primary hypotheses, we converted the RTs to measures of response cost by subtracting RTs for the new displays from those of the DTL and DTO conditions of each experiment separately. These scores were compared using an Experiment (2a vs. 2b) × Location (DTO vs. DTL) mixed ANOVA. The analysis revealed a marginally significant main effect of experiment, F(1, 52) = 3.59, p = .064, h2p = .065, and a significant main effect of location, F(1, 52) = 4.13, p = .047, h2p = .074. The interaction was not significant. Taken together, the two main effects indicated that there was greater response cost when the distractor turned target was more distant from previous target locations (in another quadrant) than when it was close to previous target locations in the learning phase. In addition, there was greater response cost associated with distractors that remained in a distractor location than with those that moved to a previous target location in the test phase.
GENERAL DISCUSSION
Figure 10. Response times (ms) of displays in the exposure and test phases of Experiment 2b. B1–B5 refer to repeated displays in Blocks 1–5 in the exposure phase. DTL (different target and location), DTO (different target only), and “new” refer to the test phase.
632
We investigated the response cost associated with previous distractors becoming targets in a variedtarget repeated search paradigm. Across four conditions, we disrupted the target–identity context association by switching target with distractor identity/location combinations or target with distractor
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
RESPONSE COST TO REPEATED DISPLAYS
Table 1. Mean reaction times (in ms) of displays in test phase of Experiments 2a and 2b Displays Experiment 2a 2b
New difference
New
DTL
DTO
DTL
DTO
1411 (175) 1439 (208)
1452 (158) 1535 (218)
1430 (172) 1496 (190)
41* (p = .026) 96** (p = .001)
19 (p = .894) 57** (p = .003)
Note: DTL = different target and location; DTO = different target only. Standard deviations and significance in parentheses. p* , .05. p** , .01.
identity only. In most conditions, participants responded significantly more slowly to repeated displays than to new displays when distractors in the repeated displays were designated as new targets in the test phase. More specifically, in Experiments 1a and 1b, the target in the test phase changed identity and location and resulted in significant response cost. In Experiments 2a and 2b, we compared targets that changed in both features to those that changed in identity but not location. Significant response cost was observed in both combinations. However, there was significantly less cost when the new target identity appeared in an old target location. In fact, in the DTO condition of Experiment 2a (same quadrant) we did not observe significant response cost. This result is addressed in detail below. Nevertheless, it appears that distractor identity and location generally contribute to the observation of response cost. From repeated exposure, participants appear to learn that some identity/ location combinations are never targets. This knowledge is expressed as slower responding to those identities/locations once they became the new targets, because the contextual or statistical past (Zellin et al., 2013) of the previous-distractor identity/location is not maintained in the test phase. Even so, the magnitude of that cost is susceptible to the influence of at least two variables: the proximity of distractors to a target during repeated search and whether the new target appeared in the same location as the old target. Distractors that were closer to an original target position generated less response cost than did distractors that were further from the target.
However, the mechanisms may be different from those of Makovski and Jiang (2010), who attributed response cost to search time from the original target position. In our study, targets were distributed throughout the display, and there was no single target position to begin a search for the new targets. Because distractors were interspersed with and near the targets in the learning phase in Experiments 1a and 2a, distractor identities were learned and associated with a specific context, although probably to a lesser degree than the target. It is reasonable to suspect that distractors near a target received some positive spillover during search (Brady & Chun, 2007). This resulted in a reduction in response cost relative to Experiments 1b and 2b in which distractors were probably learned to a lesser degree because they were more distant from any target and congregated in different regions of the displays. We also observed that changes in identity alone resulted in less cost than changes in both identity and location. It appears that having the new target appear in a previous target location may counterbalance some of the inhibition accrued for previous distractor identities. Hence, the lack of significant contextual cost in DTO condition in Experiment 2a may be due to a combination of opposing effects. There is a general response cost that we attribute to a disruption in learned identity–context associations. Cost was reduced when the distractor turned target appeared in a previous target location. As mentioned above, this cost was further reduced because of proximity between targets and distractors during repeated search. Note, however, that significant response cost
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
633
YANG AND MERRILL
was still observed when distractors were moved to previous target locations that were more distant, indicating that cost can be produced by changes in target identity alone. Thus, location information and identity information can operate independently in determining the magnitude of response cost. Taken together, our results are consistent with general models of visual search that suggest that nontarget objects can be explicitly or implicitly inhibited based on features, similarities, or spatial locations (see Crawford, Hill, & Higham, 2005; Treisman & Gelade, 1980; Wolfe, 1994). By “marking off” distractors, the capacity-limited attention system can be more efficiently deployed to possible target locations and refrain from returning to the previously visited nontarget objects (Horowitz & Wolfe, 2003). However, learning that a particular identity/location was never a target may incur a response cost when this object becomes a target in the future. Our studies indicate a complex relationship between identity inhibition, location inhibition, and vicinity of the distractors to the previous targets in producing response cost. Robust identity-based response cost was found whether location inhibition was absent or not and whether previous distractors were close to the previous target or not, but just not when both of them were operating. Original manuscript received 19 March 2014 Accepted revision received 23 December 2014 First published online 18 February 2015
REFERENCES Brady, T. F., & Chun, M. M. (2007). Spatial constraints on learning in visual search: Modeling contextual cuing. Journal Of Experimental Psychology: Human Perception And Performance, 33(4), 798–815. doi:10. 1037/0096-1523.33.4.798 Chun, M. M., & Jiang, Y. (1998). Contextual cueing: Implicit learning and memory of visual context
634
guides spatial attention. Cognitive Psychology, 36, 28–71. Conci, M., Sun, L., & Müller, H. J. (2011). Contextual remapping in visual search after predictable targetlocation changes. Psychological Research, 75(4), 279– 289. doi:10.1007/s00426-010-0306-3 Crawford, T. J., Hill, S., & Higham, S. (2005). The inhibitory effect of a recent distracter. Vision Research, 45(27), 3365–3378. doi:10.1016/j.visres. 2005.07.024 Endo, N., & Takeda, Y. (2004). Selective learning of spatial configuration and object identity in visual search. Perception & Psychophysics, 68(2), 293–302. Horowitz, T. S., & Wolfe, J. M. (2003). Memory for rejected distractors in visual search? Visual Cognition, 10(3), 257–298. doi:10.1080/ 13506280143000005 Huang, T., & Grossberg, S. (2010). Cortical dynamics of contextually cued attentive visual learning and search: Spatial and object evidence accumulation. Psychological Review, 117(4), 1080–1112. doi:10.1037/a0020664 Jiang, Y., & Song, J. (2005). Hyperspecificity in Visual Implicit Learning: Learning of Spatial Layout Is Contingent on Item Identity. Journal of Experimental Psychology: Human Perception And Performance, 31(6), 1439–1448. doi:10.1037/00961523.31.6.1439. Kunar, M. A., & Wolfe, J. M. (2011). Target absent trials in configural contextual cuing. Attention, Perception, & Psychophysics, 73(7), 2077–2091. Makovski, T., & Jiang, Y. V. (2010). Contextual cost: When a visual-search target is not where it should be. The Quarterly Journal of Experimental Psychology, 63(2), 216–225. doi:10.1080/17470210903281590 Ogawa, H., Takeda, Y., & Kumada, T. (2007). Probing attentional modulation of contextual cueing. Visual Cognition, 15, 276–289. Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention. Cognitive Psychology, 12 (1), 97–136. Wolfe, J. M. (1994). Guided search 2.0: A revised model of visual search. Psychonomic Bulletin & Review, 1(2), 202–238. doi:10.3758/BF03200774 Zellin, M., Conci, M., van Mühlenen, A., & Müller, H. J. (2013). Here today, gone tomorrow: Memory adaptation in contextual cueing. PLoS One, 8(3), e59466. doi:10.1371/journal.pone.0059466
THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2015, 68 (4)
Copyright of Quarterly Journal of Experimental Psychology is the property of Psychology Press (UK) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
