The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory

G Model

ARTICLE IN PRESS

BBR 9572 1–9

Behavioural Brain Research xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr

Research report

1

The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory

2

3

4

5 6 7

Q1

Eric A. Woodcock, Richard White, Vaibhav A. Diwadkar ∗ Department of Psychiatry and Behavioral Neurosciences, and Translational Neuroscience Program, Wayne State University School of Medicine, Detroit, MI, USA

8

9 10 11 12 13 14 15

h i g h l i g h t s • • • • •

Cognitive control is important for associative memory processes. Dorsal anterior cingulate (dACC) and dorsal prefrontal cortex (dPFC) are implicated. dACC and dPFC exert differentiable cognitive control during memory processes. dACC and dPFC exert complementary control on hippocampus during memory processes. Learning rate correlated with degree of preferential seed-target modulation.

16

17 31

a r t i c l e

i n f o

a b s t r a c t

18 19 20 21 22 23

Article history: Received 20 January 2015 Received in revised form 23 April 2015 Accepted 28 April 2015 Available online xxx

24 25 26 27 28 29 30

Keywords: Associative memory Cognitive control Psychophysiological interaction Memory encoding Memory retrieval

Cognitive control includes processes that facilitate execution of effortful cognitive tasks, including associative memory. Regions implicated in cognitive control during associative memory include the dorsal prefrontal (dPFC) and dorsal anterior cingulate cortex (dACC). Here we investigated the relative degrees of network-related interactions originating in the dPFC and dACC during oscillating phases of associative memory: encoding and cued retrieval. Volunteers completed an established object-location associative memory paradigm during fMRI. Psychophysiological interactions modeled modulatory network interactions from the dPFC and dACC during memory encoding and retrieval. Results were evaluated in second level analyses of variance with seed region and memory process as factors. Each seed exerted differentiable modulatory effects during encoding and retrieval. The dACC exhibited greater modulation (than the dPFC) on the fusiform and parahippocampal gyrus during encoding, while the dPFC exhibited greater modulation (than the dACC) on the fusiform, hippocampus, dPFC and basal ganglia. During retrieval, the dPFC exhibited greater modulation (than the dACC) in the parahippocampal gyrus, hippocampus, superior parietal lobule, and dPFC. The most notable finding was a seed by process interaction indicating that the dACC and the dPFC exerted complementary modulatory control on the hippocampus during each of the associative memory processes. These results provide evidence for differentiable, yet complementary, control-related modulation by the dACC and dPFC, while establishing the primacy of dPFC in exerting network control during both associative memory phases. Our approach and findings are relevant for understanding basic processes in human memory and psychiatric disorders that impact associative memory-related networks. © 2015 Published by Elsevier B.V.

1. Introduction

32

Q2

∗ Corresponding author at: Department of Psychiatry and Behavioral Neurosciences, Tolan Park Medical Building, Suite 5B, 3901 Chrysler Drive, Detroit, MI 48201, USA. Tel.: +1 313 577 0164; fax: +1 313 577 5900. E-mail address: [email protected] (V.A. Diwadkar).

Cognitive control is an important meta-cognitive construct encompassing a range of multidimensional processes that facilitate the execution of effortful, goal-directed cognitive tasks [1]. Cognitive control mechanisms extend across multiple cortical regions [2], and mediate processes including attentional control, selection and extraction of process-relevant information, and context-dependent

http://dx.doi.org/10.1016/j.bbr.2015.04.050 0166-4328/© 2015 Published by Elsevier B.V.

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

33 34 35 36 37 38

G Model BBR 9572 1–9 2 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

cognitive process switching [1]. Context-dependent switching is particularly relevant to cognitive control as many tasks oscillate between complementary psychological processes that may engage distinct control-related demand. Two regions of the frontal lobe are implicated in executive processes: the dorsal anterior cingulate (dACC) and the dorsal prefrontal cortex (dPFC) [3–7]. Prior functional magnetic resonance imaging (fMRI) studies have highlighted the relative specialization of these regions using tasks that require varying degrees of cognitive control. Here we extend those studies by investigating the modulatory influence of each of the dPFC and the dACC on cortical and sub-cortical targets during two complementary processes of associative memory. Specifically, we focused on assessing whether the dPFC and dACC exerted spatiallydistributed modulation on target brain regions during the encoding and retrieval of object-location associations. Memory encoding involves the pairing of distinct, often novel memoranda, such that either memoranda and/or their association can be voluntarily recalled in the future [8,9]. Effective memory encoding is dependent on several specialized regions distributed across the frontal-hippocampal network [8–12]. Germane to this investigation, the object-location association network included a diverse network of regions including the dACC, the dPFC, the parahippocampal gyrus (PHG), the fusiform gyrus, the parietal lobe, basal ganglia, and the hippocampus. The PHG is associated with spatial processing (i.e. object location) while the fusiform is associated with object recognition [8,12–14]. The basal ganglia is interconnected with the PFC and may be involved in ‘gating’ information maintained in the dPFC or object-location rehearsal during memory encoding [1,15,16]. The hippocampus is critical for the binding of memoranda (i.e. an object with its spatial location) [8,12]. Published data motivated the present investigation of the relative contributions of the two seed regions: dACC and dPFC. Evidence implicates the dACC in control of memory-relevant attention, information filtering and extraction, and conflict monitoring [1]. Specific to object–location associative memory, the dACC is hypothesized to monitor performance, ignore task-irrelevant distractions (e.g. irrelevant object-related memories or environmental stimuli, such as scanner noise), and block conflicting or competing incorrect object–location associations [1,7,8,16]. Previous fMRI studies demonstrated the dACC is active during both encoding and recognition/retrieval of associative memories [8]. However, to our knowledge, no studies have specifically assessed network profiles of the dACC during associative memory. The dPFC sub-serves the selection and active mental representation of the memoranda to be paired during encoding [11,17]. Extensive evidence implicates the dPFC’s involvement in active mental representations of memoranda during brief delay periods (i.e. working memory) [17–21]. Moreover, several studies demonstrated the dPFC’s involvement in active mental representation and organization of memoranda prior to associative memory encoding (for review see [11]). Recent work demonstrated directed interactions between the hippocampus and other constituent regions during encoding [22]. Memory binding is a process initiated by the hippocampus and consolidated through cortical–hippocampal interactions, yet, few studies have assessed directed top-down control of memory binding. Cued memory retrieval also relies on cognitive control and processing within the memory network (which implicates many of the same brain regions as encoding) [8,12]. In this study, the cue is a spatial location on a fixed three by three grid which implicates the involvement of the superior parietal lobule (SPL) and the PHG [8,12]. fMRI studies indicate an important role for the PFC in executing top-down cognitive control of these regions to facilitate accurate memory retrieval. Miller and D’Esposito demonstrated phase coherence between the PFC and the hippocampus (PFC activity preceded hippocampal activity) during memory recognition [8].

However, little is known about the unique and potentially complementary role of sub-regions within the PFC (e.g. dPFC and dACC) during these control processes. Recent work suggested that the mechanisms of action of the dPFC and the dACC occur via amplification of sub-cortical and cortical responses in process-relevant regions [23–25]. By implication, efferent signals from frontal structures may enhance cortical and sub-cortical responses during cognitive control as has been previously shown in the context of working memory [24,25]. Such evidence elaborates the putative network roles of the dACC and the dPFC: these structures are not merely specialized for processing conflict or control during tasks, but through modulation of brain networks, engage in the functional integration of control-related processing in complex cognitive tasks. The current study extends this idea and examines process-specific roles during a frontal–hippocampal associative memory paradigm [11]. 1.1. Aims and hypotheses Using a variant of a widely-employed associative memory paradigm [26], we compared the modulatory effects [27] of the dPFC and the dACC (independently) across an a priori cortical–striatal network during memory encoding and cued retrieval. The analytic approach, psychophysiological interaction (PPI), was selected to model our a priori hypotheses of distinct, and yet potentially complementary, seed-to-target effects. This analytic approach provided a robust and well-characterized linear model to assess this pattern of inter-regional and directional effective connectivity from fMRI signals [27–29]. In addition, we also conducted exploratory analyses to link behavioral proficiency (i.e., learning rate across associative memory epochs) and the relative degree of control-related modulation of the hippocampus by the dACC (>dPFC) and dPFC (>dACC) during encoding and/or retrieval. This exploration is motivated by the idea that control mechanisms act to facilitate performance (via functional integration) during relatively complex cognitive tasks.

105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120

121

122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138

2. Methods

139

2.1. Participant recruitment

140

All experimental procedures were approved by the Wayne State University Institutional Review Board. Participants provided informed consent prior to their involvement and received monetary compensation for participation. Enrolled participants (N = 10) were free of psychiatric or neurological conditions, with an average age of 22 years (range: 18–29 years; 5 females). 2.2. Imaging parameters Gradient echo fMRI signals were acquired on a 4 T system (Bruker MedSpec) using an 8 channel head coil (TR = 3 s, TA = 2 s, TE = 30 ms, matrix = 64 × 64, slices = 24, FOV = 240 mm, voxel size = 3.75 × 3.75 × 4.0 mm, images = 288). During the scan, stimuli were projected from a computer onto a screen mounted over the participant’s head. Head motion was restricted using foam inserts surrounding the participant’s head and earplugs were provided to reduce scanner noise. 2.3. Experimental design During fMRI, all participants completed the object–location associative memory paradigm. The paradigm (Fig. 1) consisted of eight memory blocks cleaved into encoding and retrieval epochs [26]. During encoding, illustrations of nine common mono-syllabic

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

141 142 143 144 145 146

147

148 149 150 151 152 153 154 155

156

157 158 159 160

G Model BBR 9572 1–9

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

3

Fig. 1. The experimental paradigm is depicted. Across eight epochs, participants viewed illustrations of nine common objects in sequential random order in each of the nine grid locations during encoding (ENC; 3 s per object; 27 s total). After a brief rest (9 s), a location was cued (RET; 3 s per object; 27 s total), and participants verbally indicated the name of the object associated with that location.

The time series from each seed region, derived from the effects of interest contrast (p < .05), was convolved with two distinct contrasts: encoding > rest and retrieval > rest. The resultant regressors represent contextual effects of the seed associated with each of encoding and retrieval. Each regressor term was positivelyweighted [27] to model the hypothesized amplification effects of cognitive control. These first-level PPI maps were submitted to a second-level random effects analysis of variance with two nonindependent factors modeled: seed (dPFC vs. dACC) and memory process (encoding vs. retrieval). Relational contrasts were used to assess main effects associated with seed and seed by process interactions. Statistical inference was based on cluster-level correction (p < .05) [35].

177

objects [3 s/object; 27 s total; [30]] were presented in sequential random order. Each object appeared in a unique location in a nine-location [3 square by 3 square] spatial grid. Each of the nine object–location pairs was consistent (unchanging) across all eight memory blocks for each participant. Following a brief rest interval [9 s], participant’s memory of each object–location pair was tested during the subsequent retrieval epoch. Locations in the grid were highlighted [black square; 3 s/square; 27 s total] and participants were required to name the associated object. Participants verbalized the object names during retrieval. Mono-syllabic object names minimized head motion associated with verbalization. Cued-recall enforced prefrontal-driven retrieval [31,32]. Participant responses were coded in the control room using the built-in microphone/speaker system. Response collection was facilitated by acquiring each image one full second before the onset of the subsequent repetition pulse (i.e., TA < TR), providing a silent window of one second within the TR envelope.

178

2.4. Image processing

To characterize the learning results, negatively accelerated learning functions (proportion correct = 1 − e−k*epoch ) were iteratively fit to block-wise performance data for each subject (0 ≤ proportion correct ≤ 1) [36] using Matlab 7.1 [37]. The single varying parameter k is a metric of learning rate (higher values represent more rapid learning) with epoch (time) as the single independent factor in the model. Across participants, negatively accelerated functions provided strong fits to the observed performance data (Mean r2 = .71, SD = .19; Mean RMSE = .10). Learning performance was robust across participants (Mean k = .73, SD = .47). Fig. 2 depicts mean performance curves over time plotted for the observed data (Fig. 2A) and estimated performance from the fitted functions (Fig. 2B).

161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176

186

Data were processed using SPM8 (Statistical Parametric Mapping, Wellcome Department of Imaging and Neuroscience, London, UK) using typical methods. Images were oriented along the AC-PC line and spatially-normalized to the MNI (Montreal Neurological Institute) template brain and smoothed by a Gaussian filter (8 mm full-width half maximum). Each epoch was modeled as a box-car vector with a canonical hemodynamic reference wave form with separate regressors for encoding, retrieval, and rest.

187

2.5. Data analysis strategy

179 180 181 182 183 184 185

188 189 190 191 192 193 194 195 196 197 198

Psychophysiological interactions (PPI) were used to model the directional modulatory effects of the dPFC and the dACC during memory encoding and retrieval. PPI forms a robust model of seedbased interactions [28,33,27] positioned between techniques of functional and maximal effective connectivity analyses [34]. PPI modeled the response of targets in terms of the interaction between a linear convolution of the physiological response of the seed and the contrast of interest (representing the psychological context). The strength of the interaction was parametrically encoded intrasubject such that it could be submitted to second-level random effects analyses of variance for group-level hypothesis testing.

199 200 201 202 203 204 205 206 207 208 209 210 211

3. Results

212

3.1. Behavioral results

213

3.2. Imaging results Results from PPI analyses are presented in the following order: modulation by each seed during each associative memory process, followed by the seed by process interaction. Finally, results assessing the statistical relationship between the magnitude of seed-to-target modulation and learning rate will be presented. Motion parameters for all images were within acceptable metrics (<2 mm). The second-level contrast of cluster-level

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

214 215 216 217 218 219 220 221 222 223 224 225 226

227

228 229 230 231 232 233 234 235

G Model

ARTICLE IN PRESS

BBR 9572 1–9

E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

4

Fig. 2. Behavioral performance is depicted as a function of time. (A) Mean observed learning performance (proportion correct) is plotted as a function of time (epoch). Mean performance was well predicted by a negatively accelerated learning function. (B) After fitting negatively accelerated performance curves to each subjects’ data (proportion correct = 1 − e−k*epoch ; see Section 3.1), the resultant data for the fitted averages are plotted providing good convergence with the observed data in (A). Error bars are ±SEM.

236 237 238 239 240 241 242 243

244 245 246 247

corrected significant activation (p < .05) for the effects of interest (both encoding and retrieval) is depicted in Fig. 3. Anatomical boundaries for each seed region (dACC [blue] and dPFC [red]) are depicted in Fig. 4. Finally, results from PPI models that contrasted the seed-based modulation (dACC > dPFC vs. dPFC > dACC) on target brain regions during each memory process (encoding vs. retrieval) are described in detail in Table 1.

3.2.1. Memory encoding results During encoding, the dACC exhibited greater modulation than the dPFC on the fusiform and parahippocampal gyrus (Fig. 5A). Similarly, during encoding, the dPFC exhibited greater modulation than

the dACC on the fusiform gyri and hippocampus (Fig. 5B). However, in contrast to the dACC, the dPFC exhibited greater modulation on the basal ganglia and dPFC (Fig. 5B). 3.2.2. Memory retrieval results During retrieval, the dPFC exhibited greater modulation than the dACC on the parahippocampal gyrus, hippocampus, superior parietal lobule, and dPFC (Fig. 6). No modulation effects survived cluster-level correction for the dACC greater than the dPFC (not pictured). 3.2.3. Seed by process interaction Seed by process interaction analyses revealed differential modulatory effects of each of the seed regions during each associative

Table 1 The significance peaks and significant cluster extents (Ke ) observed for each of the conducted contrasts (Figs. 5–7). Fig.

Contrast

Region

Ke

t value

Fig. 5A

dACC > dPFC encoding

Fig. 5B

dPFC > dACC encoding

Fig. 6

dACC > dPFC retrieval dPFC > dACC retrieval

R fusiform R PHG L fusiform R fusiform L occipital lobe R occipital lobe R hippocampus L basal ganglia R basal ganglia R dPFC L dPFC R dPFC n.s. L hippocampus L PHG L occipital lobe L SPL R dPFC L hippocampus L occipital lobe

109 58 295 141 358 541 65 655 1101 389 401 186 – 206 109 598 246 71 158 331

2.35 2.18 3.00 2.21 2.33 2.59 2.34 2.66 3.03 2.88 2.86 2.40 – 2.34 2.91 2.53 2.46 2.13 3.02 2.89

Fig. 7

Seed by process interaction A Seed by process interaction B

p voxel .011 .016 .002 .015 .011 .006 .011 .005 .002 .003 .003 .009 – .011 .002 .007 .008 .018 .001 .003

MNI (x, y, z) (34, −50, −4) (28, −46, −6) (−48, −50, −20) (36, −66, −12) (−46, −52, −14) (28, −86, −16) (26, −32, 2) (−22, 20, −4) (24, 20, −4) (30, 30, 40) (−32, 34, 36) (48, 8, 38) – (−18, −20, −18) (−28, −16, −24) (−2, −90, 28) (−22, −64, 40) (50, 2, 38) (−28, −14, −24) (−32, −90, −4)

Note: PHG = parahippcampal gyrus; dACC = dorsal anterior cingulate cortex; dPFC = dorsal prefrontal cortex; SPL = superior parietal lobule.

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

248 249 250

251 252 253 254 255 256

257 258 259

G Model BBR 9572 1–9

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

5

Fig. 4. The anatomical location and volume of each seed region is depicted. The dACC (blue) is depicted to the left and the dPFC (red) to the right. (For interpretation Q5 of the references to color in this figure legend, the reader is referred to the web version of the article.)

memory process on the hippocampus. Specifically, the dACC demonstrated greater modulation of the hippocampus during encoding, while the dPFC exhibited greater modulation during retrieval (Fig. 7). Conversely, a single cluster in primary visual cortex evinced greater modulation by the dPFC during encoding and by the dACC during retrieval (not pictured).

Fig. 3. The second-level contrast of cluster-level corrected significant activation (p < .05) for the effects of interest (both encoding and retrieval) are depicted on a mosaic of contiguous axial slices.

3.2.4. Exploratory analyses Using non-parametric tests, we conducted exploratory bivariate correlation analyses to examine the statistical relationship between the magnitude of each of the main and interaction effects observed in the modulated target regions and the observed learning rate across memory trials. Thus, we identified significant main and interaction effects (Table 1) for further exploration, with a specific focus on whether learning rate was correlated with the magnitude of preferential modulation (e.g. dPFC > dACC during encoding) in the target regions of interest (e.g. hippocampus). Parameter estimates of the degree of modulation were extracted from each

Fig. 5. (A) Clusters showing increased modulation by the dACC (relative to the dPFC) during encoding are depicted on an orthoview of the brain. The significant clusters depicted were found in the fusiform and parahippocampal gyrus. (B) Clusters showing increased modulation by the dPFC (relative to the dACC) during encoding are depicted on a mosaic of contiguous axial slices. These effects are more prevalent than the effects of the dACC (with significant clusters depicted in the hippocampus, and bilaterally in the fusiform, basal ganglia, occipital lobe and dPFC), evidence of more primary network effects exerted by the dPFC.

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

260 261 262 263 264 265

266 267 268 269 270 271 272 273 274 275 276

G Model BBR 9572 1–9 6

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

Fig. 6. Clusters showing increased modulation by the dPFC (relative to the dACC) during retrieval are depicted on a mosaic of contiguous sagittal slices. The significant clusters observed in the hippocampus, parahippocampal gyrus, occipital lobe, dPFC and superior parietal lobule suggest the preferred network signature of the dPFC during memory retrieval. 277 278 279 280 281

volume of interest (33.5 mm3 ; radius = 2 mm) centered in target peaks of significant effects and correlated (Spearman’s ) with learning rate (significance threshold; p < .05). Bivariate correlation analyses revealed three significant relationships with learning rate: dPFC (>dACC) in the fusiform (MNI: −48, −50, −20) during

encoding (Spearman’s  = .74, p < .05), dPFC (>dACC) in the primary visual cortex (MNI: −46, −52, −14) during encoding (Spearman’s  = .79, p < .05), and the seed by process interaction (dACC > dPFC during encoding; dPFC > dACC during retrieval) in the hippocampus (MNI: −28, −14, −24) (Spearman’s  = .88, p < .005). These findings reflect the statistical relationship between learning rate across memory blocks and the maximal degree of preferential seed-totarget modulation exerted in each target region. Fig. 8 presents a two-dimensional mosaic of the location of each significant correlation adjacent to a scatterplot of peak preferential modulation and learning rate. Within one continuous significant cluster (which spanned the fusiform, inferior temporal gyrus, and visual cortex) during encoding (dPFC > dACC), parameter values in two loci (fusiform gyrus and visual cortex) correlated with learning rate (the effect in the fusiform gyrus is pictured in Fig. 8A). Fig. 8B depicts the statistical relationship between the degree of peak preferential seed-to-target modulation on the hippocampus during the seed by process interaction and learning rate.

4. Discussion

Fig. 7. The depicted cluster represents the region modulated under a significant seed by process interaction. Specifically, the single significant cluster in the hippocampus represents relatively increased modulation by the dACC (relative to the dPFC) during encoding and relatively increased modulation by the dPFC (relative to the dACC) during retrieval. This effect provides the clearest evidence within our analyses of complementary network effects exerted by each of the dPFC and the dACC during periods of associative memory.

Cognitive tasks oscillate between complementary psychological processes that may engage distinct control-related demand. Here, we operationalized this inquiry and examined directed network interactions from each of the dPFC and the dACC during memory encoding and retrieval in an object-location associative memory paradigm. Our results indicate: (A) During encoding, the dACC exerted greater modulation than the dPFC on the fusiform and parahippocampal gyrus (Fig. 5A). In comparison, the dPFC exerted greater modulation than the dACC on the fusiform, hippocampus, basal ganglia and dPFC (Fig. 5B). (B) During retrieval, the dPFC exerted greater modulation than the dACC in the hippocampus, parahippocampal gyrus, and superior parietal lobule (Fig. 6). These results are succinctly summarized in Table 2. (C) A significant seed by process interaction was observed in the hippocampus. In the hippocampus, the dACC exerted greater modulation during encoding, but the dPFC exerted greater modulation during retrieval

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299

300

301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316

G Model BBR 9572 1–9

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

7

Fig. 8. Exploratory analyses investigating the relationship between behavioral and PPI-estimated connectivity metrics are depicted. Learning rate data were only available for a subset of subjects (n = 8). One individual’s data was excluded because it reflected an extreme outlier, and another’s was not available due to experimenter error. (A) Bivariate correlation analysis demonstrated learning rate (k) across the eight learning epochs was significantly associated (Spearman’s  = .74, p < .05) with the magnitude of modulation (dPFC > dACC) in the fusiform (MNI: −48, −50, −20) during encoding. (B) Bivariate correlation analysis demonstrated learning rate (k) across the eight learning epochs was significantly associated (Spearman’s  = .88, p < .005) with the magnitude of modulation (encoding: dACC > dPFC; retrieval: dPFC > dACC) in the hippocampus (MNI: −28, −14, −24).

317 318 319 320 321 322 323 324

(Fig. 7). (D) Exploratory analyses revealed significant statistical relationships between learning rate and the degree of preferential control (peak modulation) exerted by the dPFC (>dACC) during encoding in the fusiform (Fig. 8A) and the seed by process interaction (dACC > dPFC during encoding; dPFC > dACC during retrieval) in the hippocampus (Fig. 8B). These differential effects likely arise from the relatively specialized control-related roles of each of the dACC and the dPFC, and may reflect how those roles exert

Table 2 An overview of the overall design of the analyses as well as a summary of the observed results for each seed (dPFC vs. dACC) and process (encoding vs. retrieval).

effects across the memory network. In the remainder of the paper, we discuss the relevance of these findings in the context of the distinct roles of the dACC and dPFC, their implications for associative memory, and extensions to other domains. The a priori selection of the dACC and dPFC as seed regions was based on their hypothesized roles in executive processes and cognitive control [3–7]. The dACC is relatively specialized for conflict monitoring, process-relevant attentional control and information extraction; while the dPFC is implicated in active maintenance of memoranda and executive control during psychological processes, including associative memory [1,11,17]. PPI analyses facilitated investigation of how these roles might be expressed in network interactions and provided insight into directional seed-to-target relationships as a function of memory process-related demand. Thus, each result reflects the relative contribution of each seed region in exerting network control. Our findings emphasize the primacy of the dPFC during both associative memory encoding and retrieval. The preferred modulation by the dPFC during both encoding and retrieval suggested that its relatively specialized role in executive attention and memoranda maintenance exert strong network signatures during those cognitive processes [1,8,11,17]. Moreover, cued retrieval is hypothesized to originate in the dPFC, an additional process-related aspect of the structure revealed in its pattern of modulation during retrieval [11]. The role of the dACC in conflict monitoring and resolution may be reflected its network-based signatures during encoding: task-related demands during encoding include distinguishing competing memoranda (objects) from being erroneously associated with grid locations [27]. Thus, the dACC’s preferential involvement during encoding may reflect a role for monitoring and resolving conflict between competing object–location pairs that must be consolidated in longer-term memory. The precise molecular bases of these results remain slightly obscure (though see [19] for an exposition in terms of

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358

G Model BBR 9572 1–9

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

8

390

synaptic mechanisms), yet emerging evidence at the macroscopic scale suggest that the dACC and the dPFC may operate via amplification of task-relevant regions or networks [23–25]. Efferent signals from the dPFC may enhance cortical and sub-cortical responses during cognitive control. Using network-based analyses, we have previously documented evidence of this in the context of working memory [24,25]: increases in working memory load resulted in greater dACC modulation of areas in the working memory network (superior parietal lobule, dPFC, and dlPFC), that are related to task sub-goals including encoding, memoranda maintanence and active memoranda updating [25]. Those findings support the expanded role of the dACC and dPFC beyond relatively specialized conflict processing and control-related tasks, and indicated their involvement in functional integration of controlrelated processing via amplification of brain networks. The current study provides further support for, and clarification of, the processspecific roles of the dACC and dPFC via control-related amplification in the domain of associative memory. A notable effect observed in the exploratory analyses was the relationship between fMRI metrics of connectivity and behavioral proficiency in a subset of the main effects. For instance, in the fusiform/inferior temporal gyrus region, the relative increase in modulation emanating from the dPFC (compared to the dACC) correlated with learning proficiency (Fig. 8A). In addition, in the hippocampus, parameter estimates associated with the seed by process interaction (dACC > dPFC during encoding and dPFC > dACC during retrieval) also predicted learning proficiency (Fig. 8B). While speculative and incomplete, these findings suggest that connectivity metrics for some of the regions are statistically predictive of learning proficiency, providing evidence of how the amplification of process-relevant signals in target regions may mediate emergent and overt behavior.

391

4.1. Summary and conclusions

359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389

392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422

Our study provides evidence of complementary network signatures exerted by each of the dACC and the dPFC during processes of associative memory, notwithstanding methodological limitations. PPI analyses can be brittle with respect to statistical power. As such, block designs (employed in the current study) are preferred to event-related designs to improve statistical power and combat ‘temporal blurring’ associated with hemodynamic response delay [29]. Moreover, PPI models provide relatively simplistic approaches toward uncovering network interactions because they focus on pair-wise relationships, and will not capture network effects originating in seeds that were not assessed. In that sense, analyses such as ours do not provide a basis for “network discovery” [38], though this was not our goal. The a priori motivation of the current study was to compare the network signatures of the dACC and the dPFC during memory encoding and retrieval in an effort to estimate complementary control mechanisms exerted by these regions. Thus, the present findings are consistent with, and contribute to, the theorized network bases of control mechanisms. These results provide evidence that cognitive control is not unitary, but dynamic, in the sense that regional loci of control exert network effects during distinct periods of memory-related processing, specifically encoding and retrieval. In the light of recent findings [39,40], future studies are needed to investigate the relative contribution of the vPFC and dPFC during associative memory encoding and retrieval. Frontal–hippocampal mechanisms (and impairment within) have been related to a host of psychological domains and psychiatric disorders, including schizophrenia and bipolar disorder [22,26,41–46]. This approach and our results have implications not only for understanding basic processes in human memory, but also neurologic and psychiatric disorders that impact memory and memory-related brain networks (and could elucidate

compensatory mechanisms masking underlying dysfunction in individuals without overt symptomatology).

423 424

Author contributions

425

EAW was responsible for conducting the data analyses, producing the figures, and drafting the manuscript. RW guided data analyses and provided technical support. VAD collected the data presented herein, conceptualized the study design and analysis strategy, and edited the manuscript.

426 427 428 429 430

Conflict of interest

431

The authors declare no conflict of interest with respect to the conduct or content of this work.

432 433

Acknowledgements

434

This work was supported by the Elizabeth Elser Doolittle Inves- Q3 tigatorship (NARSAD), the Cohen Neuroscience Endowment, the Children’s Research Center of Michigan and the Lyckaki-Young Fund from the State of Michigan. The funding agencies played no role in the analyses or preparation of the results. References [1] Banich MT. Executive function: the search for an integrated account. Curr Direct Psychol Sci 2009;18:89–94. [2] Cole MW, Schneider W. The cognitive control network: integrated cortical regions with dissociable functions. Neuroimage 2007;37:343–60. [3] Botvinick MM, Braver TS, Barch DM, Carter CS, Cohen JD. Conflict monitoring and cognitive control. Psychol Rev 2001;108:624–52. [4] di Pellegrino G, Ciaramelli E, Ladavas E. The regulation of cognitive control following rostral anterior cingulate cortex lesion in humans. J Cognit Neurosci 2007;19:275–86. [5] Kerns JG, Cohen JD, MacDonald III AW, Cho RY, Stenger VA, Carter CS. Anterior cingulate conflict monitoring and adjustments in control. Science (New York, NY) 2004;303:1023–6. [6] Paus T. Primate anterior cingulate cortex: where motor control, drive and cognition interface. Nat Rev Neurosci 2001;2:417–24. [7] Miller EK. The prefrontal cortex and cognitive control. Nat Rev Neurosci 2000;1:59–65. [8] Miller BT, D’Esposito M. Spatial and temporal dynamics of cortical networks engaged in memory encoding and retrieval. Front Hum Neurosci 2012;6:109. [9] Jackson O, Schacter DL. Encoding activity in anterior medial temporal lobe supports subsequent associative recognition. Neuroimage 2004;21:456–62. [10] Sommer T, Rose M, Glascher J, Wolbers T, Buchel C. Dissociable contributions within the medial temporal lobe to encoding of object-location associations. Learn Memory 2005;12:343–51. [11] Simons JS, Spiers HJ. Prefrontal and medial temporal lobe interactions in longterm memory. Nat Rev Neurosci 2003;4:637–48. [12] Sommer T, Rose M, Weiller C, Buchel C. Contributions of occipital, parietal and parahippocampal cortex to encoding of object-location associations. Neuropsychologia 2005;43:732–43. [13] Epstein R, Harris A, Damian S, Kanwisher N. The parahippocampal place area: recognitition, navigation, or encoding. Neuron 1999;23:115–25. [14] Epstein R, Graham KS, Downing PE. Viewpoint-specific scene representations in human parahippocampal cortex. Neuron 2003;37:865–76. [15] O’Reilly R, Frank M. Making working memory work: a computational model of learning in the prefrontal cortex and basal ganglia. Neural Comput 2006;18:283–328. [16] Miller EK, Cohen JD. An integrative theory of prefrontal cortex function. Annu Rev Neurosci 2001;24:167–202. [17] Curtis CE, D’Esposito M. Persistent activity in the prefrontal cortex during working memory. Trends Cognit Sci 2003;7:415–23. [18] Goldman-Rakic P. Cellular basis of working memory. Neuron 1995;14:477–85. [19] Arnsten AF, Wang MJ, Paspalas CD. Neuromodulation of thought: flexibilities and vulnerabilities in prefrontal cortical network synapses. Neuron 2012;76:223–39. [20] Wang M, Yang Y, Wang C-J, Gamo NJ, Jin LE, Mazer JA, et al. NMDA receptors subserve persistent neuronal firing during working memory in dorsolateral prefrontal cortex. Neuron 2013;77:736–49. [21] Owen AM, McMillan KM, Laird AR, Bullmore E. N-back working memory paradigm: a meta-analysis of normative functional neuroimaging studies. Hum Brain Map 2005;25:46–59. [22] Wadehra S, Pruitt P, Murphy ER, Diwadkar VA. Network dysfunction during associative learning in schizophrenia: increased activation, but decreased connectivity: an fMRI study. Schizophr Res 2013;148:38–49.

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

Q4

435 436 437 438 439

440

441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492

G Model BBR 9572 1–9

ARTICLE IN PRESS E.A. Woodcock et al. / Behavioural Brain Research xxx (2015) xxx–xxx

493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522

[23] Egner T, Hirsch J. Cognitive control mechanisms resolve conflict through cortical amplification of task-relevant information. Nat Neurosci 2005;8:1784–90. [24] Bakshi N, Pruitt P, Radwan J, Keshavan MS, Rajan U, Zajac-Benitez C, et al. Inefficiently increased anterior cingulate modulation of cortical systems during working memory in young offspring of schizophrenia patients. J Psychiatric Res 2011;45:1067–76. [25] Banyai M, Diwadkar VA, Erdi P. Model-based dynamical analysis of functional disconnection in schizophrenia. Neuroimage 2011;58:870–7. [26] Diwadkar VA, Flaugher B, Jones T, Zalanyi L, Ujfalussy B, Keshavan MS, et al. Impaired associative learning in schizophrenia: behavioral and computational studies. Cognit Neurodyn 2008;2:207–19. [27] Friston KJ, Buechel C, Fink GR, Morris J, Rolls E, Dolan RJ. Psychophysiological and modulatory interactions in neuroimaging. Neuroimage 1997;6:218–29. [28] Friston KJ. Functional and effective connectivity: a review. Brain Connect 2011;1:13–36. [29] O’Reilly JX, Woolrich MW, Behrens TE, Smith SM, Johansen-Berg H. Tools of the trade: psychophysiological interactions and functional connectivity. Soc Cogn Affect Neurosci 2012;7:604–9. [30] Snodgrass JG, Vanderwart M. A standardized set of 260 pictures: norms for name agreement, image agreement, familiarity, and visual complexity. J Exp Psychol Hum Learn Memory 1980;6:174–215. [31] Allan K, Dolan RJ, Fletcher PC, Rugg MD. The role of the right anterior prefrontal cortex in episodic retrieval. Neuroimage 2000;11:217–27. [32] Desgranges B, Baron JC, Eustache F. The functional neuroanatomy of episodic memory: the role of the frontal lobes, the hippocampal formation, and other areas. Neuroimage 1998;8:198–213. [33] Kim J, Horwitz B. Investigating the neural basis for fMRI-based functional connectivity in a blocked design: application to interregional correlations and psycho-physiological interactions. Magn Reson Imaging 2008;26:583–93. [34] Stephan KE. On the role of general system theory for functional neuroimaging. J Anat 2004;205:443–70.

9

[35] Ward BD. Simultaneous inference for fMRI data. Milwaukee, WI: Medical College of Wisconsin; 2000. [36] Heathcote A, Brown S, Mewhort DJ. The power law repealed: the case for an exponential law of practice. Psychon Bull Rev 2000;7:185–207. [37] MathWorks I. Matlab. 7.1 ed. Natick, MA: The MathWorks, Inc.; 2007. [38] Friston KJ, Li B, Daunizeau J, Stephan KE. Network discovery with DCM. Neuroimage 2012;56:1202–21. [39] Barredo J, Öztekin I, Badre D. Ventral fronto-temporal pathway supporting cognitive control of episodic memory retrieval. Cerebral Cortex 2013:bht291. [40] Bunge S, Burrows B, Wagner A. Prefrontal and hippocampal contributions to visual associative recognition: interactions between cognitive control and episodic retrieval. Brain Cognit 2004;56:141–52. [41] Érdi P, Flaugher B, Jones T, Ujfalussy B, Zalányi L, Diwadkar VA. Computational approach to the schizophrenia: disconnection syndrome and dynamical pharmacology. In: Ricciardi LM, Buonocore A, Pirozzi E, editors. BIOCOMP. Meville, NY: American Institute of Physics; 2007. p. 65–87. [42] Brambilla P, Cerruti S, Bellani M, Ferro A, Marinelli V, Giusto D, et al. Shared impairment in associative learning in schizophrenia and bipolar disorder. Prog Neuro-psychopharmacol Biol Psychiatry 2011;35:1093–9. [43] Brambilla P, Riva MA, Melcangi R, Diwadkar VA. The role of glutamatergic pathways in schizophrenia: from animal models to human imaging studies. Clin Neuropsychiatry 2007;4:199–207. [44] Heckers S. Neuroimaging studies of the hippocampus in schizophrenia. Hippocampus 2001;11:520–8. [45] Hasselmo ME. Neuromodulation and the hippocampus: memory function and dysfunction in a network simulation. Prog Brain Res 1999;121:3–18. [46] Siekmeier PJ, Hasselmo ME, Howard MW, Coyle J. Modeling of contextdependent retrieval in hippocampal region CA1: implications for cognitive function in schizophrenia. Schizophr Res 2007;89:177–90.

Please cite this article in press as: Woodcock EA, et al. The dorsal prefrontal and dorsal anterior cingulate cortices exert complementary network signatures during encoding and retrieval in associative memory. Behav Brain Res (2015), http://dx.doi.org/10.1016/j.bbr.2015.04.050

523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551