Neuroscience Letters 581 (2014) 52–56
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Comparison of associative recognition versus source recognition Heekyeong Park ∗ , Cheryl Abellanoza, James D. Schaeffer University of Texas at Arlington, United States
h i g h l i g h t s • • • •
Object pairs were encoded with two study contexts. MTL activity was found for associative and source recognition. Cortical activity differed between associative and source recognition. MTL is involved in concurrent retrieval of associative and source memories.
a r t i c l e
i n f o
Article history: Received 20 May 2014 Received in revised form 18 July 2014 Accepted 13 August 2014 Available online 23 August 2014 Keywords: fMRI Associative memory Source memory Hippocampus MTL
a b s t r a c t The importance of the medial temporal lobe (MTL) for memory of arbitrary associations has been well established. However, the contribution of the MTL in concurrent retrieval of different classes of associations remains unclear. The present fMRI study investigated neural correlates of concurrent retrieval of associative and source memories. Participants studied a list of object pairs with two study tasks and judged the status and context of the pair during test. Associative retrieval was supported by neural activity in bilateral prefrontal cortex and left ventral occipito-temporal cortex, while source recognition was linked to activity in the right caudate. Both the hippocampus and MTL cortex showed retrieval activity for associative and source memory. Importantly, greater brain activity for successful associative recognition accompanied with successful source recognition was evident in left perirhinal and anterior hippocampal regions. These results indicate that the MTL is critical in the retrieval of different classes of associations. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction According to dual process models of recognition, episodic memory can be supported by recollection of contextual details of an event or by acontextual familiarity-based recognition of the event [1,2]. The contribution of these processes in recognition has often been investigated by comparing correct versus incorrect judgment of the item-item association (associative memory henceforth) [3,4] or correct versus incorrect judgment of the item-context association (source memory hereafter) [5,6]. Previous fMRI studies have indicated the distinction between recollection and familiarity by uncovering a recollection network where memory judgments accompanied by recollection tend to be disproportionally correlated with activity in the medial temporal lobe (MTL), posterior parietal cortex, and medial prefrontal cortex during retrieval [7,8].
∗ Corresponding author: Department of Psychology, University of Texas at Arlington, Arlington, TX 75069-19528, United States. Tel.: +1 817 272 1063; fax: +1 817 272 2364. E-mail address: [email protected] (H. Park). http://dx.doi.org/10.1016/j.neulet.2014.08.024 0304-3940/© 2014 Elsevier Ireland Ltd. All rights reserved.
Considering that both associative and source memories are supported by recollection from the perspective of the dual process account, it is reasonable to ask if a single neural network underlies both memories for associations. Influential theoretical accounts of recognition postulate equivalent memory representations and mechanisms for different associations [9,10]. Although previous fMRI studies have reported different cortical activity for associative and source memory during encoding [11,12], it is still plausible that both memories are engaged in the same recollection network for retrieval of a cohesive representation of arbitrary associations. It is currently unknown whether associative and source recognition would show dissociations during retrieval, as found with encoding. In order to examine whether associative memory and source memory require independent neural activity in any extent during retrieval, the direct comparison of retrieval between two memories could be informative. The aims of the present study were to address (1) whether concurrent retrieval of associative and source memories for an event would be supported by activity in the single neural network for recollection, and/or (2) whether associative or source recognition would employ different neural activity for its own retrieval
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
mechanism. As such, the present study compared retrieval of associative and source memories for a single event at the whole brain level. Participants studied a list of picture pairs depicting objects with two study tasks. Participants were then administered associative recognition (association of objects in the pair) and source recognition (association of the pair and study task) tests. With this procedure, we investigated the neural network for concurrent associative and source recognition.
53
paired with different items at study with the fit judgment, (iv) Rearranged/Common: two studied items that had been paired with different items at study with the common judgment, or (v) New: two unstudied items. Participants were instructed to respond with ‘New’ if they were unsure about the pair or task, or if they recognized only one item of the pair. Both study and test phases were held in the scanner. 2.4. fMRI scanning
2. Methods 2.1. Participants Twenty volunteers participated in the experiment (18–27 years; 10 males). They were right-handed, native English speakers who reported no history of neurological or psychiatric illness. Volunteers gave informed consent prior to participation, and they were compensated for their participation. The experiment was approved by the University of Texas at Arlington and the University of Texas Southwestern Medical Center Institutional Review Boards. One participant was excluded from analysis due to incomplete data.
A 3 T MR scanner equipped with a 32-channel head coil was used to acquire both T1 -weighted high-resolution anatomical images (MP-RAGE, 240 × 240 matrix, 1 mm3 , sagittal) and T∗2 -weighted echo-planar images (flip angle 70◦ , 80 × 80 matrix, FOV 24 cm, TR 2000 ms, TE 30 ms, SENSE factor 1.5) per volume. Each volume comprised 33 slices oriented parallel to the AC-PC line (3 mm3 , 1 mm gap, axial) acquired in a descending sequence. Imaging data from the test phase were acquired in two scan sessions comprising 340 volumes each. Five additional volumes were collected at the beginning of each session but were discarded to allow for T1 equilibration. The 3.7 s SOA allowed an effective sampling rate of the hemodynamic response of approximately 2 Hz.
2.2. Materials 2.5. fMRI data analysis The stimuli were drawn from a pool of 383 picture pairs depicting objects (Supplementary Methods). A study list comprised a pseudorandom ordering of 270 unrelated picture pairs such that each pair appeared equally often in each task context and each item in a pair was presented equally often to the left and right of a fixation cross. The study list also contained 90 perceptual baseline trials consisting of an arrow pointing either to the left or right. A test list comprised 360 pairs. Among them, 180 were studied pairs presented in the same pairing as at study (intact pairs), 90 pairs comprised studied items from the same study task but had been re-paired from study (rearranged pairs), and 90 were new pairs. Both study and test lists were constrained such that no pair from the same task occurred more than three times consecutively and were separately constructed for each subject. An additional 23 pairs were used for practice. 2.3. Procedure Participants were given instructions and practice prior to the experiment proper. The experiment consisted of a single studytest cycle. On each study trial, a white cross appeared on the screen (200 ms), followed by a task cue (500 ms) indicating the task to be performed on the upcoming pair: ‘FIT’ or ‘COMMON’ (Supplementary Figure 1). For the ‘FIT’ task cue, participants judged whether an item in the pair fit into the other item in the pair. For the ‘COMMON’ task cue, participants judged whether the two items could be put together perceptually (e.g., Can they be found in the same place?, Can they be in the same color?, etc.). A pair was then displayed for 3 s, with one picture presented to the left and the other to the right of the fixation cross. Participants indicated their corresponding task judgments by pressing a button using their right hand. For perceptual baseline trials, an arrow pointing either to the left or the right was displayed for 1.6 s, and participants were asked to press the button in the opposite direction of the arrowhead. The test was administered approximately 3–5 min after the end of the study phase. For each test trial, a picture pair was displayed for 3.7 s. Participants were instructed to make one of five associative/source recognition responses indicating the status of the pair and the study task: (i) Intact/Fit: two items studied in the same pairing as study with the fit judgment, (ii) Intact/Common: two items studied in the same pairing as study with the common judgment, (iii) Rearranged/Fit: two studied items that had been
Statistical Parametric Mapping (SPM8) was used for data preprocessing and statistical analyses. For each participant, functional images were spatially realigned to the mean image, time-corrected to the middle slice, reoriented, normalized to the MNI EPI template, and smoothed with an 8 mm full-width half-maximum Gaussian kernel. Each participant’s anatomical scan was normalized to the MNI T1 template and averaged to create an across-subjects anatomical mean image. Statistical analysis was performed on the test phase data using a two-stage mixed effects model. Prior to model estimation, image time series were concatenated across runs. In the first stage, stimulus-elicited neural activity was modeled with 2 s boxcar functions. The event-related blood oxygen-level dependent (BOLD) response was modeled by convolving these boxcar functions with a canonical hemodynamic response function (HRF). In addition, six regressors were employed to model movement-related variance along with session regressors. Parameter estimates for events of interest were estimated for each subject using a General Linear Model. Non-sphericity of the error covariance was accommodated by an AR(1) model [13]. Linear contrasts were constructed for each subject and entered into second level tests. For analysis of concurrent retrieval of associative and source memory, four events of interest were defined: ‘associative hits-source hits’ (accurately judged intact pairs with correct study task judgment), ‘associative hits-source misses’ (accurately judged intact pairs albeit with incorrect task judgment), ‘associative misses-source hits’ (intact pairs inaccurately identified as rearranged but with correct task judgment), and ‘associative misses-source misses’ (intact pairs inaccurately identified as rearranged with incorrect task judgment). Item misses and recognition judgments to rearranged and new pairs were separately modeled but not further analyzed. Events of no interest included trials with omitted or multiple responses. For the analysis at the whole-brain level, a repeated measures 2 × 2 ANOVA (memory: associative vs. source, accuracy: hit vs. miss) was implemented in SPM to identify neural activity for retrieval of associative memory and source memory. Effects were thresholded at p < .001, uncorrected with a 5 voxel extent threshold. For significant interaction effects, pair-wise t-tests were conducted for follow-up tests on parameter estimates extracted from peak voxels to examine the pattern of differences contributing
54
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
Table 1 Brain regions exhibiting activity for associative recognition and source recognition. Coordinates (x y z) Associative recognition (associative hit > associative miss) −2 −51 29 54 29 −2 39 −25 64 −1 16 −42 41 25 −6 −16 28 0 −48 −70 22 57 −22 16 −54 −76 7 −9 −49 −2 −85 4 −15 −55 −44 3 Source recognition (source hit > source miss) 12 8 −14
Z
Cluster size
5.73 4.05 4.51 3.61 3.47 4.39 4.43 3.54 3.51 3.78 6.04 3.50
272 22 57 5 6 267 278 7 6 12 603 6
3.17
28
to the interaction identified by the aforementioned ANOVA. Note that the interaction from the ANOVA is non-directional. In light of the importance of MTL in associative and source recognition, a complementary ROI analysis was performed using an inclusive masking procedure with a manually drawn MTL mask. Clusters were accepted as significant if a cluster consisted of at least 5 contiguous voxels exceeding the threshold of p < .005 after inclusive masking. The significance level for follow-up statistical tests was set to p < .05. 3. Results 3.1. Behavioral results Proportions and reaction times (RTs) of recognition judgments are displayed in Supplementary Table 1. In order to see the effect of conjoint associative and source recognition analogous to the functional analysis, accuracy of recognition to intact test pairs was analyzed with a 2 × 2 repeated measures ANOVA with factors of memory (associative, source) and recognition accuracy (hit, miss). This analysis revealed main effects of associative recognition (F[1,18] = 13.06, p < .005) and source recognition (F[1,18] = 9.69, p < .01), indicating that participants made more correct judgments on both associative and source recognition. There was a marginal interaction between associative recognition and source recognition (F[1,18] = 3.75. p = .05) such that correct source judgments were more likely to be made when associative judgments were also accurately made (t[18] = 6.68, p < .001). The discriminability index for accurate associative recognition (d ) for correct source judgments (1.00 [.11]) was greater than d for incorrect source judgments (.45 [.07], t[18] = 6.52, p < .001), although both were significantly above chance (ps < .001). Further, RT data showed that recognition judgments were made faster for correct judgments than for incorrect judgments for both associative memory (F[1,18] = 13.06, p = .005) and source memory (F[1,18] = 9.68, p = .01). However, there was no interaction effect in RTs between memory type and recognition accuracy. 3.2. fMRI results 3.2.1. Main effects of associative and source retrieval In the 2 × 2 ANOVA model at the whole brain level, we first sought main effects of associative retrieval and source retrieval separately. Table 1 lists the outcome of these analyses. In order to find brain regions for associative memory retrieval where neural activity for correctly recognized intact test pairs differed from activity for intact pairs incorrectly judged as rearranged pairs, we exclusively masked the main effect of associative recognition with the interaction effect of associative and source memory to eliminate
Region
BA
L ventrolateral prefrontal cortex R ventrolateral prefrontal cortex R precentral gyrus (posterior part) L rolandic operculum L anterior cingulate cortex Mid-cingulate cortex L mid temporal cortex R superior temporal gyrus L mid occipital cortex L lingual gyrus L occipital cortex/lingual gyrus R cerebellum
47/45 45 6/4 48 32 24/23 39/37 42 19 18 17/18/19/30
R caudate
25
brain regions in which associative retrieval differed by source retrieval. We used an analogous procedure to find neural correlates of source recognition by searching for brain regions where recognition activity of pairs with correct source judgments differed from recognition activity of pairs without correct source judgments. The source retrieval contrast was then exclusively masked with the interaction of associative and source recognition to remove brain regions where source memory retrieval differed by associative recognition. At the cortical level, activity for associative recognition (associative hit > associative miss) was found notably in bilateral prefrontal areas including the ventrolateral prefrontal cortex, the precentral gyrus, and anterior to mid cingulate cortex. In addition, associative recognition was linked to activity in extensive regions encompassing the left temporal to occipital cortex. For source memory, only a single suprathreshold cluster was identified in the right caudate, where retrieval activity of source hits was greater than retrieval activity of source misses. At the MTL level, a single cluster was identified as the main effect of associative recognition in the left hippocampus (−36, −22, −17). Source retrieval activity was also found in the right hippocampus extending to entorhinal cortex (39, −22, −14). Parameter estimates extracted from these MTL clusters demonstrated only the main effect of interest with no alternative effect or interactions (Supplementary Figure 2).
3.2.2. Interaction of associative and source memories At the cortical level, interaction effects of associative and source recognition were shown in the left putamen, right anterior cingulate cortex, and mid-frontal gyrus (Supplementary Figure 3). In all three regions, activity for associative hits was greater than activity for associative misses only when accompanied by accurate source recognition. In contrast, the difference between activity for associative hits versus associative misses was either negligible (anterior cingulate cortex) or in the opposite direction (left putamen, right mid-frontal gyrus) when source retrieval was not successful.1 Further, two left MTL clusters, one in perirhinal cortex and the other in the anterior hippocampus, revealed the same pattern of greater retrieval activity for associative hits than associative misses only with accurate source retrieval (Fig. 1). However, no overlapped activity for retrieval of associative and source memory was found either at the cortical or MTL level.2
1 Simple effect tests yielded indistinguishable results. We thank Reviewer 1 for the suggestion. 2 We also ran the analysis with anatomical masks to see interactions independently and found almost identical results.
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
Fig. 1. MTL regions that showed the interaction between associative and source recognition and parameter estimates from peak voxels of clusters (AH, associative hit; AM, associative miss; SH, source hit; SM, source miss).
4. Discussion The main goal of the present study was to investigate neural activity for concurrent retrieval of associative and source recognition. The current results demonstrate that, similar to encoding, associative and source retrieval rely on at least partially different cortical mechanisms. Associative retrieval was supported by activity in bilateral ventrolateral prefrontal cortex, the right precentral gyrus, and mid cingulate cortex, along with activity spanning extensive regions from the ventral occipito-temporal cortex. On the other hand, successful source recognition was related to activity in the right caudate. Behaviorally, an interaction between associative and source recognition was expressed as accuracy in associative recognition accompanied by source recognition. Analogously, interaction effects identified in the left putamen, right anterior cingulate cortex, and mid frontal gyrus revealed greater activity for successful associative recognition only with successful source retrieval. At the MTL level, the left hippocampus was linked to associative recognition while the right hippocampus extending to entorhinal cortex was related to source recognition. The pattern of greater brain activity for successful associative recognition accompanied with successful source recognition was also evident in left perirhinal and anterior hippocampal clusters. We did not find overlapped activity for both associative and source recognition; however, the influence of interaction activity in MTL and other cortical areas was evident. We will detail the functional significance of these results. The involvement of prefrontal cortex in associative memory is consistent with prior findings that have demonstrated the importance of ventrolateral prefrontal cortex (LVPFC) in controlled retrieval (anterior) and selection processes (mid) [14,15]. Given that retrieval of associations between two items seems to require not only controlled retrieval of semantic knowledge but also goaldriven selection processes, the current finding of VLPFC seems to support the two-process model of VLPFC [16]. Further, there was significantly greater mid cingulate cortex for associative hits compared to associative misses in the present study. Simultaneous reactivation of the original pair of items and blocked reactivation of other items that had been studied would most likely to be critical for successful associative retrieval. Under such circumstances,
55
increased neural response in the mid cingulate cortex indicated the possibility that cognitive conflict resolution or the demand for cognitive control could be an important indicator of successful associative recognition [17,18]. In addition, retrieval activity in the occipito-temporal cortex tends to reflect reactivation of perceptual features of visually studied items; this is in line with the proposal that successful retrieval of an episode depends on the reactivation of cortical patterns that occurred during encoding [19]. Further, retrieval of source information elicited activity in the caudate. The role of the caudate in memory has been suggested in previous imaging studies, particularly regarding spatial working memory [20,21] and feedback processing to the response and outcome for successful task performance [22,23]. In the present study, participants were asked to retrieve the study task context in which a pair of objects was studied for source recognition. Reactivating visuospatial features of representations of studied pairs in working memory seems to be necessary for determining the context of the studied pair. In addition, evaluating an accurate combination of the study task and the test pair could be dependent upon not only reactivating/retaining relevant study information but also filtering out irrelevant information. In that sense, caudate activity seems to reflect monitoring of the match between the test pair and study context by taking only relevant context information while filtering irrelevant context information through the inspection of visuospatial details of recovered memories as an estimation of appropriateness. For the interaction between associative and source memory, the pattern of greater activity for associative hits accompanied by source hits was observed in the left putamen, right anterior cingulate cortex, and right middle frontal gyrus. The putamen is well known for its influence in various types of learning and memory, and it is a part of the basal ganglia, a neural gatekeeper [24,25]. Moreover, the anterior cingulate cortex has been considered as the critical region for conflict detection and cognitive control [18,26], while the right middle frontal region has been reported for its relation to episodic retrieval [27,28]. Importantly, all three regions have been associated with cognitive control processes such as inhibiting irrelevant information (R putmen & anterior cingulate cortex) and maintaining relevant information (R middle frontal cortex) [29,30]. Therefore, the current results seem to suggest that functional contributions of these regions are rooted in monitoring inaccurate associations (i.e., rearranged pair, mismatch between the test pair and contexts) while reactivating accurate associations (i.e., intact pair, match between the test pair and study context) as a key process for successful retrieval of both associative and source memory. In concordance with prior findings of the MTL involvement in associative and source recognition [31,32], we also found that the hippocampus and entorhinal cortex were linked to associative and source recognition, respectively. Notably, left perirhinal and anterior hippocampal regions showed correlated activity patterns between associative and source recognition such that recognition of both associative and source memory elicited greater activity. These findings implicate that associative recognition and source recognition are correlated at some extent in the MTL, along with cortical interaction effects. Although the finding of perirhinal activity does not exclude the contribution of familiarity-based recognition entirely [33,34], perirhinal activity could reflect activity for object associations [35–37] and object discrimination [38–40], as object pairs were involved in both associative and source recognition, and the discrimination of the studied object pairs with studied context from other pairs was likely crucial for successful retrieval. Taken together, the current findings indicate the involvement of MTL in associative and source memory through interaction effects during retrieval.
56
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
5. Conclusion The present study compared retrieval of associative and source memory. The current findings demonstrated that associative and source recognition recruited activity from different cortical regions. At the same time, both associative and source recognition relied on activity in MTL regions. This indicated the possibility that associative and source recognition are based on different neural correlates at the cortical level but still require the MTL as the common neural mechanism for retrieval of associative and source information. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.neulet.2014.08.024. References [1] G. Mandler, Recognizing: the judgment of previous occurrence, Psychol. Rev. 87 (1980) 252–271. [2] A.P. Yonelinas, The nature of recollection and familiarity: a review of 30 years of research, J. Mem. Lang. 46 (2002) 441–517. [3] S.E. Prince, S.M. Daselaar, R. Cabeza, Neural correlates of relational memory: successful encoding and retrieval of semantic and perceptual associations, J. Neurosci. 25 (2005) 1203–1210. [4] K.S. Giovanello, D. Schnyer, M. Verfallie, Distinct hippocampal regions make unique contribution to relational memory, Hippocampus 19 (2009) 111–117. [5] S. Cansino, P. Maquet, R.J. Dolan, M.D. Rugg, Brain activity underlying encoding and retrieval of source memory, Cereb. Cortex. 12 (2002) 1048–1056. [6] A. Duarte, R.N.A. Henson, K.S. Graham, The effect of stimulus content on the neural correlates of source recollection, Brain Res. 1373 (2011) 110–123. [7] J. Spaniol, P.S.R. Davidson, A.S.N. Kim, H. Han, M. Moscovitch, C.L. Grady, Eventrelated fMRI studies of episodic encoding and retrieval: meta-analyses using activation likelihood estimation, Neuropsychologia 47 (2009) 1765–1779. [8] H. Kim, Dissociating the roles of the default-mode, dorsal, and ventral networks in episodic memory retrieval, Neuroimage 50 (2010) 1648–1657. [9] M.W. Brown, J.P. Aggleton, Recognition memory: what are the roles of the perirhinal cortex and hippocampus? Nat. Rev. Neurosci. 2 (2001) 51–61. [10] H. Eichenbaum, A.P. Yonelinas, C. Ranganath, The medial temporal lobe and recognition memory, Annu. Rev. Neurosci. 30 (2007) 123–152. [11] H. Park, V. Shannon, J. Biggan, C. Spann, Neural activity supporting the formation of associative memory versus source memory, Brain Res. 1471 (2012) 81–92. [12] J.X. Wong, M. de Chastelaine, M.D. Rugg, Comparison of the neural correlates of encoding item–item and item–context associations, Front. Hum. Neurosci. 7 (2013), http://dx.doi.org/10.3389/fnhum.2013.00436. [13] K.J. Friston, D.E. Glaser, R.N. Henson, S. Kiebel, C. Phillips, J. Ashburner, Classical and Bayesian inference in neuroimaging: application, Neuroimage 16 (2002) 484–512. [14] D. Badre, R.A. Poldrack, E.J. Paré-Blagoev, R.Z. Insler, A.D. Wagner, Dissociable controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal cortex, Neuron 47 (2005) 907–918. [15] B.T. Gold, D.A. Balota, S.J. Jones, D.K. Powell, C.D. Smith, A.H. Andersen, Dissociation of automatic and strategic lexical-semantics: functional magnetic resonance imaging evidence for differing roles of multiple frontotemporal regions, J. Neurosci. 26 (2006) 6523–6532. [16] D. Badre, A.D. Wagner, Left ventrolateral prefrontal cortex and the cognitive control of memory, Neuropsychologia 45 (2007) 2883–2901.
[17] J.M. Fincham, J.R. Anderson, Distinct roles of the anterior cingulate and prefrontal cortex in acquisition and performance of a cognitive skill, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 12941–12946. [18] B.A. Kuhl, I. Kahn, N.M. Dudukovic, A.D. Wagner, Overcoming suppression in order to remember: contributions from anterior cingulate and ventrolateral prefrontal cortex, Cogn. Affect. Behav. Neurosci. 8 (2008) 211–221. [19] M.D. Rugg, J.D. Johnson, H. Park, M. Uncapher, Encoding-retrieval overlap in human episodic memory: a functional neuroimaging prospective, Prog. Brain Res. 169 (2008) 339–352. [20] A. Belger, A. Puce, J.H. Krystal, J.C. Gore, P. Goldman-Rakic, G. McCarthy, Dissociation of mnemonic and perceptual processes during spatial and nonspatial working memory using fMRI, Hum. Brain Mapp. 6 (1998) 14–32. [21] B.R. Postle, M. D’Esposito, Dissociation of human caudate nucleus activity in spatial and nonspatial working memory, Cogn. Brain Res. 8 (1999) 107–115. [22] J.P. O’Doherty, P. Dayan, J. Schulz, R. Deichmann, K. Friston, R.J. Dolan, Dissociable roles of ventral and dorsal striatum in instrumental conditioning, Science 304 (2004) 452–454. [23] C.A. Seger, C.M. Cincotta, The roles of the caudate nucleus in human classification learning, J. Neurosci. 16 (2005) 2941–2951. [24] B. Baier, H.O. Karnath, M. Dieterich, F. Birklein, C. Heinze, N.G. Müller, Keeping memory clear and stable – the contribution of human basal gangalia and prefrontal cortex to working memory, J. Neurosci. 30 (2010) 9788–9792. [25] F. McNab, T. Klingberg, Prefrontal cortex and basal ganglia control access to working memory, Nat. Neurosci. 11 (2008) 103–107. [26] M.H. Sohn, M.V. Albert, K. Jung, C.S. Carter, J.R. Anderson, Anticipation of conflict monitoring in the anterior cingulated cortex and the prefrontal cortex, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 10330–10334. [27] R.L. Buckner, A.Z. Snyder, A.L. Sanders, M.E. Raichle, J.C. Morris, Functional brain imaging of young, nondemented, and demented older adults, J. Cogn. Neurosci. 12 (2000) 24–34. [28] I.G. Dobbins, H.J. Rice, A.D. Wagner, D.L. Schacter, Memory orientation and success: separable neurocognitive components underlying episodic recognition, Neuropsychologia 41 (2003) 318–333. [29] M.P. Milham, M.T. Banich, Anterior cingulate cortex: an fMRI analysis of conflict specificity & functional differentiation, Hum. Brain Mapp. 25 (2005) 328–335. [30] M.N. Rajah, R. Languay, C.L. Grady, Age-related changes in right middle frontal gyrus volume correlate with altered episodic retrieval activity, J. Neurosci. 31 (2011) 17941–17954. [31] A.P. Yonelinas, L.J. Otten, K.N. Shaw, M.D. Rugg, Separating the brain regions involved in recollection and familiarity in recognition memory, J. Neurosci. 25 (2005) 3002–3008. [32] I. Tendolkar, J. Arnold, K.M. Petersson, S. Weis, A. Brockhaus-Demke, P. van Eijndhoven, J. Buitelaar, G. Gernandex, Contributions of the medial temporal lobe to declarative memory retrieval: manipulating the amount of contextual retrieval, Learn. Mem. 15 (2008) 611–617. [33] A.L. Haskins, A.P. Yonelinas, J.R. Quamme, C. Ranganath, Perirhinal cortex supports encoding and familiarity-based recognition of novel associations, Neuron 59 (2008) 554–560. [34] D. Montaldi, A.R. Mayes, The role of recollection and familiarity in the functional differentiation of the medial temporal lobes, Hippocampus 20 (2010) 1291–1314. [35] E.A. Murray, B.J. Richmond, Role of perirhinal cortex in object perception, memory, and associations, Curr. Opin. Neurobiol. 11 (2001) 188–193. [36] B.P. Staresina, L. Davachi, Object unitization and associative memory formation are supported by distinct brain regions, J. Neurosci. 30 (2010) 9890–9897. [37] H.C. Watson, E.L. Widling, K.S. Graham, A role for perirhinal cortex in memory for novel object-context associations, J. Neurosci. 32 (2012) 4473–4481. [38] L. Ryan, J.A. Cardoza, M.D. Barense, K.H. Kawa, J. Wallentin-Flores, W.T. Arnold, G.E. Alexander, Age-related impairment in a complex object discrimination task that engages perirhinal cortex, Hippocampus 22 (2012) 1978–1989. [39] A. Clarke, L.K. Tyler, Object-specific semantic coding in human perirhinal cortex, J. Neurosci. 34 (2014) 4766–4775. [40] H.C. Watson, A.C.H. Lee, The perirhinal cortex and recognition memory interference, J. Neurosci. 33 (2013) 4192–4200.
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Comparison of associative recognition versus source recognition Heekyeong Park ∗ , Cheryl Abellanoza, James D. Schaeffer University of Texas at Arlington, United States
h i g h l i g h t s • • • •
Object pairs were encoded with two study contexts. MTL activity was found for associative and source recognition. Cortical activity differed between associative and source recognition. MTL is involved in concurrent retrieval of associative and source memories.
a r t i c l e
i n f o
Article history: Received 20 May 2014 Received in revised form 18 July 2014 Accepted 13 August 2014 Available online 23 August 2014 Keywords: fMRI Associative memory Source memory Hippocampus MTL
a b s t r a c t The importance of the medial temporal lobe (MTL) for memory of arbitrary associations has been well established. However, the contribution of the MTL in concurrent retrieval of different classes of associations remains unclear. The present fMRI study investigated neural correlates of concurrent retrieval of associative and source memories. Participants studied a list of object pairs with two study tasks and judged the status and context of the pair during test. Associative retrieval was supported by neural activity in bilateral prefrontal cortex and left ventral occipito-temporal cortex, while source recognition was linked to activity in the right caudate. Both the hippocampus and MTL cortex showed retrieval activity for associative and source memory. Importantly, greater brain activity for successful associative recognition accompanied with successful source recognition was evident in left perirhinal and anterior hippocampal regions. These results indicate that the MTL is critical in the retrieval of different classes of associations. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction According to dual process models of recognition, episodic memory can be supported by recollection of contextual details of an event or by acontextual familiarity-based recognition of the event [1,2]. The contribution of these processes in recognition has often been investigated by comparing correct versus incorrect judgment of the item-item association (associative memory henceforth) [3,4] or correct versus incorrect judgment of the item-context association (source memory hereafter) [5,6]. Previous fMRI studies have indicated the distinction between recollection and familiarity by uncovering a recollection network where memory judgments accompanied by recollection tend to be disproportionally correlated with activity in the medial temporal lobe (MTL), posterior parietal cortex, and medial prefrontal cortex during retrieval [7,8].
∗ Corresponding author: Department of Psychology, University of Texas at Arlington, Arlington, TX 75069-19528, United States. Tel.: +1 817 272 1063; fax: +1 817 272 2364. E-mail address: [email protected] (H. Park). http://dx.doi.org/10.1016/j.neulet.2014.08.024 0304-3940/© 2014 Elsevier Ireland Ltd. All rights reserved.
Considering that both associative and source memories are supported by recollection from the perspective of the dual process account, it is reasonable to ask if a single neural network underlies both memories for associations. Influential theoretical accounts of recognition postulate equivalent memory representations and mechanisms for different associations [9,10]. Although previous fMRI studies have reported different cortical activity for associative and source memory during encoding [11,12], it is still plausible that both memories are engaged in the same recollection network for retrieval of a cohesive representation of arbitrary associations. It is currently unknown whether associative and source recognition would show dissociations during retrieval, as found with encoding. In order to examine whether associative memory and source memory require independent neural activity in any extent during retrieval, the direct comparison of retrieval between two memories could be informative. The aims of the present study were to address (1) whether concurrent retrieval of associative and source memories for an event would be supported by activity in the single neural network for recollection, and/or (2) whether associative or source recognition would employ different neural activity for its own retrieval
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
mechanism. As such, the present study compared retrieval of associative and source memories for a single event at the whole brain level. Participants studied a list of picture pairs depicting objects with two study tasks. Participants were then administered associative recognition (association of objects in the pair) and source recognition (association of the pair and study task) tests. With this procedure, we investigated the neural network for concurrent associative and source recognition.
53
paired with different items at study with the fit judgment, (iv) Rearranged/Common: two studied items that had been paired with different items at study with the common judgment, or (v) New: two unstudied items. Participants were instructed to respond with ‘New’ if they were unsure about the pair or task, or if they recognized only one item of the pair. Both study and test phases were held in the scanner. 2.4. fMRI scanning
2. Methods 2.1. Participants Twenty volunteers participated in the experiment (18–27 years; 10 males). They were right-handed, native English speakers who reported no history of neurological or psychiatric illness. Volunteers gave informed consent prior to participation, and they were compensated for their participation. The experiment was approved by the University of Texas at Arlington and the University of Texas Southwestern Medical Center Institutional Review Boards. One participant was excluded from analysis due to incomplete data.
A 3 T MR scanner equipped with a 32-channel head coil was used to acquire both T1 -weighted high-resolution anatomical images (MP-RAGE, 240 × 240 matrix, 1 mm3 , sagittal) and T∗2 -weighted echo-planar images (flip angle 70◦ , 80 × 80 matrix, FOV 24 cm, TR 2000 ms, TE 30 ms, SENSE factor 1.5) per volume. Each volume comprised 33 slices oriented parallel to the AC-PC line (3 mm3 , 1 mm gap, axial) acquired in a descending sequence. Imaging data from the test phase were acquired in two scan sessions comprising 340 volumes each. Five additional volumes were collected at the beginning of each session but were discarded to allow for T1 equilibration. The 3.7 s SOA allowed an effective sampling rate of the hemodynamic response of approximately 2 Hz.
2.2. Materials 2.5. fMRI data analysis The stimuli were drawn from a pool of 383 picture pairs depicting objects (Supplementary Methods). A study list comprised a pseudorandom ordering of 270 unrelated picture pairs such that each pair appeared equally often in each task context and each item in a pair was presented equally often to the left and right of a fixation cross. The study list also contained 90 perceptual baseline trials consisting of an arrow pointing either to the left or right. A test list comprised 360 pairs. Among them, 180 were studied pairs presented in the same pairing as at study (intact pairs), 90 pairs comprised studied items from the same study task but had been re-paired from study (rearranged pairs), and 90 were new pairs. Both study and test lists were constrained such that no pair from the same task occurred more than three times consecutively and were separately constructed for each subject. An additional 23 pairs were used for practice. 2.3. Procedure Participants were given instructions and practice prior to the experiment proper. The experiment consisted of a single studytest cycle. On each study trial, a white cross appeared on the screen (200 ms), followed by a task cue (500 ms) indicating the task to be performed on the upcoming pair: ‘FIT’ or ‘COMMON’ (Supplementary Figure 1). For the ‘FIT’ task cue, participants judged whether an item in the pair fit into the other item in the pair. For the ‘COMMON’ task cue, participants judged whether the two items could be put together perceptually (e.g., Can they be found in the same place?, Can they be in the same color?, etc.). A pair was then displayed for 3 s, with one picture presented to the left and the other to the right of the fixation cross. Participants indicated their corresponding task judgments by pressing a button using their right hand. For perceptual baseline trials, an arrow pointing either to the left or the right was displayed for 1.6 s, and participants were asked to press the button in the opposite direction of the arrowhead. The test was administered approximately 3–5 min after the end of the study phase. For each test trial, a picture pair was displayed for 3.7 s. Participants were instructed to make one of five associative/source recognition responses indicating the status of the pair and the study task: (i) Intact/Fit: two items studied in the same pairing as study with the fit judgment, (ii) Intact/Common: two items studied in the same pairing as study with the common judgment, (iii) Rearranged/Fit: two studied items that had been
Statistical Parametric Mapping (SPM8) was used for data preprocessing and statistical analyses. For each participant, functional images were spatially realigned to the mean image, time-corrected to the middle slice, reoriented, normalized to the MNI EPI template, and smoothed with an 8 mm full-width half-maximum Gaussian kernel. Each participant’s anatomical scan was normalized to the MNI T1 template and averaged to create an across-subjects anatomical mean image. Statistical analysis was performed on the test phase data using a two-stage mixed effects model. Prior to model estimation, image time series were concatenated across runs. In the first stage, stimulus-elicited neural activity was modeled with 2 s boxcar functions. The event-related blood oxygen-level dependent (BOLD) response was modeled by convolving these boxcar functions with a canonical hemodynamic response function (HRF). In addition, six regressors were employed to model movement-related variance along with session regressors. Parameter estimates for events of interest were estimated for each subject using a General Linear Model. Non-sphericity of the error covariance was accommodated by an AR(1) model [13]. Linear contrasts were constructed for each subject and entered into second level tests. For analysis of concurrent retrieval of associative and source memory, four events of interest were defined: ‘associative hits-source hits’ (accurately judged intact pairs with correct study task judgment), ‘associative hits-source misses’ (accurately judged intact pairs albeit with incorrect task judgment), ‘associative misses-source hits’ (intact pairs inaccurately identified as rearranged but with correct task judgment), and ‘associative misses-source misses’ (intact pairs inaccurately identified as rearranged with incorrect task judgment). Item misses and recognition judgments to rearranged and new pairs were separately modeled but not further analyzed. Events of no interest included trials with omitted or multiple responses. For the analysis at the whole-brain level, a repeated measures 2 × 2 ANOVA (memory: associative vs. source, accuracy: hit vs. miss) was implemented in SPM to identify neural activity for retrieval of associative memory and source memory. Effects were thresholded at p < .001, uncorrected with a 5 voxel extent threshold. For significant interaction effects, pair-wise t-tests were conducted for follow-up tests on parameter estimates extracted from peak voxels to examine the pattern of differences contributing
54
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
Table 1 Brain regions exhibiting activity for associative recognition and source recognition. Coordinates (x y z) Associative recognition (associative hit > associative miss) −2 −51 29 54 29 −2 39 −25 64 −1 16 −42 41 25 −6 −16 28 0 −48 −70 22 57 −22 16 −54 −76 7 −9 −49 −2 −85 4 −15 −55 −44 3 Source recognition (source hit > source miss) 12 8 −14
Z
Cluster size
5.73 4.05 4.51 3.61 3.47 4.39 4.43 3.54 3.51 3.78 6.04 3.50
272 22 57 5 6 267 278 7 6 12 603 6
3.17
28
to the interaction identified by the aforementioned ANOVA. Note that the interaction from the ANOVA is non-directional. In light of the importance of MTL in associative and source recognition, a complementary ROI analysis was performed using an inclusive masking procedure with a manually drawn MTL mask. Clusters were accepted as significant if a cluster consisted of at least 5 contiguous voxels exceeding the threshold of p < .005 after inclusive masking. The significance level for follow-up statistical tests was set to p < .05. 3. Results 3.1. Behavioral results Proportions and reaction times (RTs) of recognition judgments are displayed in Supplementary Table 1. In order to see the effect of conjoint associative and source recognition analogous to the functional analysis, accuracy of recognition to intact test pairs was analyzed with a 2 × 2 repeated measures ANOVA with factors of memory (associative, source) and recognition accuracy (hit, miss). This analysis revealed main effects of associative recognition (F[1,18] = 13.06, p < .005) and source recognition (F[1,18] = 9.69, p < .01), indicating that participants made more correct judgments on both associative and source recognition. There was a marginal interaction between associative recognition and source recognition (F[1,18] = 3.75. p = .05) such that correct source judgments were more likely to be made when associative judgments were also accurately made (t[18] = 6.68, p < .001). The discriminability index for accurate associative recognition (d ) for correct source judgments (1.00 [.11]) was greater than d for incorrect source judgments (.45 [.07], t[18] = 6.52, p < .001), although both were significantly above chance (ps < .001). Further, RT data showed that recognition judgments were made faster for correct judgments than for incorrect judgments for both associative memory (F[1,18] = 13.06, p = .005) and source memory (F[1,18] = 9.68, p = .01). However, there was no interaction effect in RTs between memory type and recognition accuracy. 3.2. fMRI results 3.2.1. Main effects of associative and source retrieval In the 2 × 2 ANOVA model at the whole brain level, we first sought main effects of associative retrieval and source retrieval separately. Table 1 lists the outcome of these analyses. In order to find brain regions for associative memory retrieval where neural activity for correctly recognized intact test pairs differed from activity for intact pairs incorrectly judged as rearranged pairs, we exclusively masked the main effect of associative recognition with the interaction effect of associative and source memory to eliminate
Region
BA
L ventrolateral prefrontal cortex R ventrolateral prefrontal cortex R precentral gyrus (posterior part) L rolandic operculum L anterior cingulate cortex Mid-cingulate cortex L mid temporal cortex R superior temporal gyrus L mid occipital cortex L lingual gyrus L occipital cortex/lingual gyrus R cerebellum
47/45 45 6/4 48 32 24/23 39/37 42 19 18 17/18/19/30
R caudate
25
brain regions in which associative retrieval differed by source retrieval. We used an analogous procedure to find neural correlates of source recognition by searching for brain regions where recognition activity of pairs with correct source judgments differed from recognition activity of pairs without correct source judgments. The source retrieval contrast was then exclusively masked with the interaction of associative and source recognition to remove brain regions where source memory retrieval differed by associative recognition. At the cortical level, activity for associative recognition (associative hit > associative miss) was found notably in bilateral prefrontal areas including the ventrolateral prefrontal cortex, the precentral gyrus, and anterior to mid cingulate cortex. In addition, associative recognition was linked to activity in extensive regions encompassing the left temporal to occipital cortex. For source memory, only a single suprathreshold cluster was identified in the right caudate, where retrieval activity of source hits was greater than retrieval activity of source misses. At the MTL level, a single cluster was identified as the main effect of associative recognition in the left hippocampus (−36, −22, −17). Source retrieval activity was also found in the right hippocampus extending to entorhinal cortex (39, −22, −14). Parameter estimates extracted from these MTL clusters demonstrated only the main effect of interest with no alternative effect or interactions (Supplementary Figure 2).
3.2.2. Interaction of associative and source memories At the cortical level, interaction effects of associative and source recognition were shown in the left putamen, right anterior cingulate cortex, and mid-frontal gyrus (Supplementary Figure 3). In all three regions, activity for associative hits was greater than activity for associative misses only when accompanied by accurate source recognition. In contrast, the difference between activity for associative hits versus associative misses was either negligible (anterior cingulate cortex) or in the opposite direction (left putamen, right mid-frontal gyrus) when source retrieval was not successful.1 Further, two left MTL clusters, one in perirhinal cortex and the other in the anterior hippocampus, revealed the same pattern of greater retrieval activity for associative hits than associative misses only with accurate source retrieval (Fig. 1). However, no overlapped activity for retrieval of associative and source memory was found either at the cortical or MTL level.2
1 Simple effect tests yielded indistinguishable results. We thank Reviewer 1 for the suggestion. 2 We also ran the analysis with anatomical masks to see interactions independently and found almost identical results.
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
Fig. 1. MTL regions that showed the interaction between associative and source recognition and parameter estimates from peak voxels of clusters (AH, associative hit; AM, associative miss; SH, source hit; SM, source miss).
4. Discussion The main goal of the present study was to investigate neural activity for concurrent retrieval of associative and source recognition. The current results demonstrate that, similar to encoding, associative and source retrieval rely on at least partially different cortical mechanisms. Associative retrieval was supported by activity in bilateral ventrolateral prefrontal cortex, the right precentral gyrus, and mid cingulate cortex, along with activity spanning extensive regions from the ventral occipito-temporal cortex. On the other hand, successful source recognition was related to activity in the right caudate. Behaviorally, an interaction between associative and source recognition was expressed as accuracy in associative recognition accompanied by source recognition. Analogously, interaction effects identified in the left putamen, right anterior cingulate cortex, and mid frontal gyrus revealed greater activity for successful associative recognition only with successful source retrieval. At the MTL level, the left hippocampus was linked to associative recognition while the right hippocampus extending to entorhinal cortex was related to source recognition. The pattern of greater brain activity for successful associative recognition accompanied with successful source recognition was also evident in left perirhinal and anterior hippocampal clusters. We did not find overlapped activity for both associative and source recognition; however, the influence of interaction activity in MTL and other cortical areas was evident. We will detail the functional significance of these results. The involvement of prefrontal cortex in associative memory is consistent with prior findings that have demonstrated the importance of ventrolateral prefrontal cortex (LVPFC) in controlled retrieval (anterior) and selection processes (mid) [14,15]. Given that retrieval of associations between two items seems to require not only controlled retrieval of semantic knowledge but also goaldriven selection processes, the current finding of VLPFC seems to support the two-process model of VLPFC [16]. Further, there was significantly greater mid cingulate cortex for associative hits compared to associative misses in the present study. Simultaneous reactivation of the original pair of items and blocked reactivation of other items that had been studied would most likely to be critical for successful associative retrieval. Under such circumstances,
55
increased neural response in the mid cingulate cortex indicated the possibility that cognitive conflict resolution or the demand for cognitive control could be an important indicator of successful associative recognition [17,18]. In addition, retrieval activity in the occipito-temporal cortex tends to reflect reactivation of perceptual features of visually studied items; this is in line with the proposal that successful retrieval of an episode depends on the reactivation of cortical patterns that occurred during encoding [19]. Further, retrieval of source information elicited activity in the caudate. The role of the caudate in memory has been suggested in previous imaging studies, particularly regarding spatial working memory [20,21] and feedback processing to the response and outcome for successful task performance [22,23]. In the present study, participants were asked to retrieve the study task context in which a pair of objects was studied for source recognition. Reactivating visuospatial features of representations of studied pairs in working memory seems to be necessary for determining the context of the studied pair. In addition, evaluating an accurate combination of the study task and the test pair could be dependent upon not only reactivating/retaining relevant study information but also filtering out irrelevant information. In that sense, caudate activity seems to reflect monitoring of the match between the test pair and study context by taking only relevant context information while filtering irrelevant context information through the inspection of visuospatial details of recovered memories as an estimation of appropriateness. For the interaction between associative and source memory, the pattern of greater activity for associative hits accompanied by source hits was observed in the left putamen, right anterior cingulate cortex, and right middle frontal gyrus. The putamen is well known for its influence in various types of learning and memory, and it is a part of the basal ganglia, a neural gatekeeper [24,25]. Moreover, the anterior cingulate cortex has been considered as the critical region for conflict detection and cognitive control [18,26], while the right middle frontal region has been reported for its relation to episodic retrieval [27,28]. Importantly, all three regions have been associated with cognitive control processes such as inhibiting irrelevant information (R putmen & anterior cingulate cortex) and maintaining relevant information (R middle frontal cortex) [29,30]. Therefore, the current results seem to suggest that functional contributions of these regions are rooted in monitoring inaccurate associations (i.e., rearranged pair, mismatch between the test pair and contexts) while reactivating accurate associations (i.e., intact pair, match between the test pair and study context) as a key process for successful retrieval of both associative and source memory. In concordance with prior findings of the MTL involvement in associative and source recognition [31,32], we also found that the hippocampus and entorhinal cortex were linked to associative and source recognition, respectively. Notably, left perirhinal and anterior hippocampal regions showed correlated activity patterns between associative and source recognition such that recognition of both associative and source memory elicited greater activity. These findings implicate that associative recognition and source recognition are correlated at some extent in the MTL, along with cortical interaction effects. Although the finding of perirhinal activity does not exclude the contribution of familiarity-based recognition entirely [33,34], perirhinal activity could reflect activity for object associations [35–37] and object discrimination [38–40], as object pairs were involved in both associative and source recognition, and the discrimination of the studied object pairs with studied context from other pairs was likely crucial for successful retrieval. Taken together, the current findings indicate the involvement of MTL in associative and source memory through interaction effects during retrieval.
56
H. Park et al. / Neuroscience Letters 581 (2014) 52–56
5. Conclusion The present study compared retrieval of associative and source memory. The current findings demonstrated that associative and source recognition recruited activity from different cortical regions. At the same time, both associative and source recognition relied on activity in MTL regions. This indicated the possibility that associative and source recognition are based on different neural correlates at the cortical level but still require the MTL as the common neural mechanism for retrieval of associative and source information. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.neulet.2014.08.024. References [1] G. Mandler, Recognizing: the judgment of previous occurrence, Psychol. Rev. 87 (1980) 252–271. [2] A.P. Yonelinas, The nature of recollection and familiarity: a review of 30 years of research, J. Mem. Lang. 46 (2002) 441–517. [3] S.E. Prince, S.M. Daselaar, R. Cabeza, Neural correlates of relational memory: successful encoding and retrieval of semantic and perceptual associations, J. Neurosci. 25 (2005) 1203–1210. [4] K.S. Giovanello, D. Schnyer, M. Verfallie, Distinct hippocampal regions make unique contribution to relational memory, Hippocampus 19 (2009) 111–117. [5] S. Cansino, P. Maquet, R.J. Dolan, M.D. Rugg, Brain activity underlying encoding and retrieval of source memory, Cereb. Cortex. 12 (2002) 1048–1056. [6] A. Duarte, R.N.A. Henson, K.S. Graham, The effect of stimulus content on the neural correlates of source recollection, Brain Res. 1373 (2011) 110–123. [7] J. Spaniol, P.S.R. Davidson, A.S.N. Kim, H. Han, M. Moscovitch, C.L. Grady, Eventrelated fMRI studies of episodic encoding and retrieval: meta-analyses using activation likelihood estimation, Neuropsychologia 47 (2009) 1765–1779. [8] H. Kim, Dissociating the roles of the default-mode, dorsal, and ventral networks in episodic memory retrieval, Neuroimage 50 (2010) 1648–1657. [9] M.W. Brown, J.P. Aggleton, Recognition memory: what are the roles of the perirhinal cortex and hippocampus? Nat. Rev. Neurosci. 2 (2001) 51–61. [10] H. Eichenbaum, A.P. Yonelinas, C. Ranganath, The medial temporal lobe and recognition memory, Annu. Rev. Neurosci. 30 (2007) 123–152. [11] H. Park, V. Shannon, J. Biggan, C. Spann, Neural activity supporting the formation of associative memory versus source memory, Brain Res. 1471 (2012) 81–92. [12] J.X. Wong, M. de Chastelaine, M.D. Rugg, Comparison of the neural correlates of encoding item–item and item–context associations, Front. Hum. Neurosci. 7 (2013), http://dx.doi.org/10.3389/fnhum.2013.00436. [13] K.J. Friston, D.E. Glaser, R.N. Henson, S. Kiebel, C. Phillips, J. Ashburner, Classical and Bayesian inference in neuroimaging: application, Neuroimage 16 (2002) 484–512. [14] D. Badre, R.A. Poldrack, E.J. Paré-Blagoev, R.Z. Insler, A.D. Wagner, Dissociable controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal cortex, Neuron 47 (2005) 907–918. [15] B.T. Gold, D.A. Balota, S.J. Jones, D.K. Powell, C.D. Smith, A.H. Andersen, Dissociation of automatic and strategic lexical-semantics: functional magnetic resonance imaging evidence for differing roles of multiple frontotemporal regions, J. Neurosci. 26 (2006) 6523–6532. [16] D. Badre, A.D. Wagner, Left ventrolateral prefrontal cortex and the cognitive control of memory, Neuropsychologia 45 (2007) 2883–2901.
[17] J.M. Fincham, J.R. Anderson, Distinct roles of the anterior cingulate and prefrontal cortex in acquisition and performance of a cognitive skill, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 12941–12946. [18] B.A. Kuhl, I. Kahn, N.M. Dudukovic, A.D. Wagner, Overcoming suppression in order to remember: contributions from anterior cingulate and ventrolateral prefrontal cortex, Cogn. Affect. Behav. Neurosci. 8 (2008) 211–221. [19] M.D. Rugg, J.D. Johnson, H. Park, M. Uncapher, Encoding-retrieval overlap in human episodic memory: a functional neuroimaging prospective, Prog. Brain Res. 169 (2008) 339–352. [20] A. Belger, A. Puce, J.H. Krystal, J.C. Gore, P. Goldman-Rakic, G. McCarthy, Dissociation of mnemonic and perceptual processes during spatial and nonspatial working memory using fMRI, Hum. Brain Mapp. 6 (1998) 14–32. [21] B.R. Postle, M. D’Esposito, Dissociation of human caudate nucleus activity in spatial and nonspatial working memory, Cogn. Brain Res. 8 (1999) 107–115. [22] J.P. O’Doherty, P. Dayan, J. Schulz, R. Deichmann, K. Friston, R.J. Dolan, Dissociable roles of ventral and dorsal striatum in instrumental conditioning, Science 304 (2004) 452–454. [23] C.A. Seger, C.M. Cincotta, The roles of the caudate nucleus in human classification learning, J. Neurosci. 16 (2005) 2941–2951. [24] B. Baier, H.O. Karnath, M. Dieterich, F. Birklein, C. Heinze, N.G. Müller, Keeping memory clear and stable – the contribution of human basal gangalia and prefrontal cortex to working memory, J. Neurosci. 30 (2010) 9788–9792. [25] F. McNab, T. Klingberg, Prefrontal cortex and basal ganglia control access to working memory, Nat. Neurosci. 11 (2008) 103–107. [26] M.H. Sohn, M.V. Albert, K. Jung, C.S. Carter, J.R. Anderson, Anticipation of conflict monitoring in the anterior cingulated cortex and the prefrontal cortex, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 10330–10334. [27] R.L. Buckner, A.Z. Snyder, A.L. Sanders, M.E. Raichle, J.C. Morris, Functional brain imaging of young, nondemented, and demented older adults, J. Cogn. Neurosci. 12 (2000) 24–34. [28] I.G. Dobbins, H.J. Rice, A.D. Wagner, D.L. Schacter, Memory orientation and success: separable neurocognitive components underlying episodic recognition, Neuropsychologia 41 (2003) 318–333. [29] M.P. Milham, M.T. Banich, Anterior cingulate cortex: an fMRI analysis of conflict specificity & functional differentiation, Hum. Brain Mapp. 25 (2005) 328–335. [30] M.N. Rajah, R. Languay, C.L. Grady, Age-related changes in right middle frontal gyrus volume correlate with altered episodic retrieval activity, J. Neurosci. 31 (2011) 17941–17954. [31] A.P. Yonelinas, L.J. Otten, K.N. Shaw, M.D. Rugg, Separating the brain regions involved in recollection and familiarity in recognition memory, J. Neurosci. 25 (2005) 3002–3008. [32] I. Tendolkar, J. Arnold, K.M. Petersson, S. Weis, A. Brockhaus-Demke, P. van Eijndhoven, J. Buitelaar, G. Gernandex, Contributions of the medial temporal lobe to declarative memory retrieval: manipulating the amount of contextual retrieval, Learn. Mem. 15 (2008) 611–617. [33] A.L. Haskins, A.P. Yonelinas, J.R. Quamme, C. Ranganath, Perirhinal cortex supports encoding and familiarity-based recognition of novel associations, Neuron 59 (2008) 554–560. [34] D. Montaldi, A.R. Mayes, The role of recollection and familiarity in the functional differentiation of the medial temporal lobes, Hippocampus 20 (2010) 1291–1314. [35] E.A. Murray, B.J. Richmond, Role of perirhinal cortex in object perception, memory, and associations, Curr. Opin. Neurobiol. 11 (2001) 188–193. [36] B.P. Staresina, L. Davachi, Object unitization and associative memory formation are supported by distinct brain regions, J. Neurosci. 30 (2010) 9890–9897. [37] H.C. Watson, E.L. Widling, K.S. Graham, A role for perirhinal cortex in memory for novel object-context associations, J. Neurosci. 32 (2012) 4473–4481. [38] L. Ryan, J.A. Cardoza, M.D. Barense, K.H. Kawa, J. Wallentin-Flores, W.T. Arnold, G.E. Alexander, Age-related impairment in a complex object discrimination task that engages perirhinal cortex, Hippocampus 22 (2012) 1978–1989. [39] A. Clarke, L.K. Tyler, Object-specific semantic coding in human perirhinal cortex, J. Neurosci. 34 (2014) 4766–4775. [40] H.C. Watson, A.C.H. Lee, The perirhinal cortex and recognition memory interference, J. Neurosci. 33 (2013) 4192–4200.
