Neural processing of recollection, familiarity and priming at encoding: Evidence from a forced-choice recognition paradigm

brain research 1585 (2014) 72–82

Available online at www.sciencedirect.com

www.elsevier.com/locate/brainres

Research Report

Neural processing of recollection, familiarity and priming at encoding: Evidence from a forced-choice recognition paradigm Yingfang Menga,n, Xiaohong Yeb, Brian D. Gonsalvesc a

Department of Psychology, Fujian Normal University, China School of Education and Music, Sanming University, China c Department of Psychology, California State University, East Bay, USA b

ar t ic l e in f o

abs tra ct

Article history:

The distinction between neural mechanisms of explicit and implicit expressions of memory has

Accepted 11 August 2014

been well studied at the retrieval stage, but less at encoding. In addition, dissociations obtained

Available online 16 August 2014

in many studies are complicated by methodological difficulties in obtaining process-pure

Keywords:

measures of different types of memory. In this experiment, we applied a subsequent memory

Recollection

paradigm and a two-stage forced-choice recognition test to classify study ERP data into four

Familiarity

categories: subsequent remembered (later retrieved accompanied by detailed information),

Priming

subsequent known (later retrieved accompanied by a feeling of familiarity), subsequent primed

Encoding

(later retrieved without conscious awareness) and subsequent forgotten (not retrieved).

DM effect

Differences in subsequent memory effects (DM effects) were measured by comparing ERP waveform associated with later memory based on recollection, familiarity or priming with ERP waveform for later forgotten items. The recollection DM effect involved a robust sustained (onset at 300 ms) prefrontal positive-going DM effect which was right-lateralized, and a later (onset at 800 ms) occipital negative-going DM effect. Familiarity involved an earlier (300–400 ms) prefrontal positive-going DM effect and a later (500–600 ms) parietal positive-going DM effect. Priming involved a negative-going DM effect which onset at 600 ms, mainly distributed over anterior brain sites. These results revealed a sequence of components that represented cognitive processes underlying the encoding of verbal information into episodic memory, and separately supported later remembering, knowing and priming. & 2014 Elsevier B.V. All rights reserved.

1.

Introduction

On any given day, we encounter and experience many events. Only some of these experiences are transformed into memories and can be subsequently remembered. One n

Corresponding author. E-mail address: [email protected] (Y. Meng).

http://dx.doi.org/10.1016/j.brainres.2014.08.024 0006-8993/& 2014 Elsevier B.V. All rights reserved.

outcome of encoding can be explicit memory, which is an expression of memory involving the conscious awareness of prior events, and accompanies voluntary retrieval of studied information. Explicit memory has usually been examined by using intentional tests of recall and recognition memory.

brain research 1585 (2014) 72–82

In such tests of explicit remembering, experiences that cannot be retrieved intentionally are labeled as subsequently forgotten. However, in some indirect or implicit tests that present previously studied information in a seemingly unrelated task (e.g., perceptual identification), a recent encounter with an item can be shown to influence responding even in the absence of explicit memory. This phenomenon is called implicit memory, or priming, and refers to a long-term change in behavioral response to an item as a result of prior exposure to it, usually taking the form of facilitated processing. Much evidence from event-related potentials (ERPs) and functional magnetic resonance imaging (fMRI) is consistent with the idea that explicit memory and implicit memory have different neural bases at retrieval. For example, ERPs have revealed different spatiotemporal components associated with implicit and explicit retrieval (Rugg et al., 1998; Paller et al., 2003; Henson et al., 2003; Meng and Guo, 2006; review also see Dew and Cabeza, 2011). fMRI studies have reported reduced neural responses for primed relative to novel stimuli, primarily in the occipitotemporal cortices during implicit retrieval, and enhanced neural responses in medial temporal lobe, prefrontal cortex and posterior medial parietal cortex for explicit retrieval (Henson, 2003; Schott et al., 2005). These results imply that implicit access to memory is supported by neural processing that is qualitatively distinct from that supporting conscious memory access. While many studies have observed dissociations between neural correlates of implicit and explicit memory at retrieval, very few studies have explored similar dissociations at encoding. The ability to remember a past event is not only influenced by processes at retrieval, but also depends on processes engaged at the time of event encoding. A powerful method of examining the neural basis of successful encoding is to measure neural activity during the study phase of an experiment and then sort these measurements according to subsequent memory test performance. Sanquist et al. (1980) first observed that the ERPs elicited by study items that were subsequently remembered elicited larger positive amplitudes over midline parietal scalp sites than those that were subsequently forgotten. Paller et al. (1987) labeled this parietalmaximal neural signature as the “difference in subsequent memory effect” (DM effect). This DM effect has been replicated in many studies using intentional tests, and has been taken as an indicator of encoding process yielding later explicit memory (Friedman and Trott, 2000; Schott et al., 2002; Guo et al., 2005). Although its timing and topography vary depending upon the precise experimental conditions, typical DM effects tend to have either a frontal scalp distribution or a posterior scalp distribution, which has been attributed to neural processing in the prefrontal cortex (PFC) and medial temporal lobes (MTL) respectively (Donaldson and Rugg, 1999; Buckner et al., 2000; Duarte et al., 2004). Evidence from neuroimaging studies using fMRI and PET has consistently shown that PFC and MTL activation is associated with the encoding of new information into explicit memory (Simons and Spiers, 2003; for a review, Cabeza and Moscovitch, 2013). PFC is assumed to be involved in processing and organization of incoming information (elaborative processing), interacting with the MTL to effect memory storage (Moscovitch, 1992).

73

However, DM effects reflecting encoding process yielding later implicit memory have not been consistently observed by using implicit tests, and where such effects have been observed, they have tended to resemble explicit memory DM effects (Friedman et al., 1996; Paller et al., 1987; Paller, 1990, for a review, see Schott et al., 2002). One difficulty in interpreting the extant data is the methodological issue that neural substrates of implicit and explicit memory are predominantly identified using very different specialized memory tests. This is problematic, given strong evidence that explicit and implicit memory tests are not necessarily process-pure (Voss and Paller, 2008). In other words, on any given memory test given to a healthy individual, performance may be guided by both explicit memory processes and implicit memory processes. A new methodological approach that can concurrently capture the operation of distinct memory processes for specific episodes in a single memory test is thus essential. Schott et al., (2002) first employed a novel paradigm to measure explicit memory and priming-without-explicit memory in one test, and contrasted the neural signals of these two processes at the encoding stage via a DM analysis. They used deep (semantic) versus shallow (non-semantic) encoding conditions, followed by a two-stage procedure in which threeletter word stems were presented in the first stage, and participants attempted to complete each stem with a word from the preceding study list. If they could not, they completed the stem with the first word that came to mind. After each stem was completed, subjects indicated whether or not they recognized the word from the encoding phase. Some of the test stems could be completed with study list words, and these stems were divided into three critical trial types: stems completed with studied words and indicated to be from the study list were termed “remembered” trials; stems completed with studied words but indicated not to be from the study list were termed “primed” trials; and stems completed with unstudied words and indicated not to be from the study list were termed “forgotten” trials, that is, the corresponding studied words had been forgotten, thus this trial type forming a suitable baseline. The three types of test trials were used to classify ERPs recorded during the study phase. The DM effect for explicit memory was evaluated by contrasting study ERPs corresponding to remembered and forgotten test trials. This DM effect was similar to that observed in previous studies in which ERPs corresponding to subsequently remembered trials elicited larger positive amplitudes than those corresponding to subsequently forgotten trials. The effect was observed over right frontal sites in the 900–1200 ms time window regardless of level of processing, and central scalp sites from 600 to 800 ms during shallow study processing only. On the other hand, the DM effect for priming-without-explicit memory was evaluated by contrasting study ERPs corresponding to primed and forgotten test items. This DM was distinct from the DM effect associated with explicit memory, that is, ERPs corresponding to subsequently primed trials elicited larger negative amplitudes than those corresponding to subsequently forgotten, and this DM was observed over centroparietal scalp locations from 200 to 450 ms. These results showed for the first time that implicit and explicit memory have distinct neural correlates at encoding.

74

brain research 1585 (2014) 72–82

Meng (2012, 2013) adopted a similar two-stage testing procedure to that employed by Schott et al. (2002) to investigate the negative-going DM effect for priming. ERPs were recorded while participants studied visually presented words (Meng, 2012) or novel faces (Meng, 2013), followed by a twostage forced-choice recognition test where two items (one studied and one new) were presented concurrently, and subjects were asked to choose the studied one. If subjects could not recognize the studied item, guessing was permitted. After choosing, subjects reported whether their selection had been made based on recognition or guessing. The DM for explicit memory was identified by contrasting study ERPs to items which were later selected in the forced-choice recognition test and recognized correctly as studied (remembered trials) versus study ERPs to items which were not later selected in the forced-choice recognition test (forgotten trials), that is, the new items were selected as studied in the forced-choice recognition test instead. The DM for implicit memory was identified by contrasting study ERPs to items which were later selected in the forced-choice recognition test but were selected by a guess (primed trials) versus study ERPs to forgotten items. According to Voss et al. (2008) and Voss and Paller (2009a), implicit memory can contribute to forced-choice recognition in the absence of explicit memory. Studied items may be selected in forced-choice tests because perceptual fluency is systematically greater than that of the accompanying novel items, even in the absence of explicit memory. Meng found that, with word stimuli (Meng, 2012), implicit memory was associated with a negative-going DM effect beginning 200 ms after stimulus onset, maximal at temporal scalp sites, whereas explicit memory was associated with a right prefrontal, positive-going DM effect in 400–600 ms. With face stimuli (Meng, 2013), implicit memory was associated with a negative DM effect over frontal–central electrodes in 400–500 ms, and explicit memory was associated with a positive DM effect over parietal electrodes starting 400 ms after face onset. These two studies using a similar paradigm with different stimulus types both showed distinct neural signals associated with implicit and explicit memory at encoding. Additional evidence for the dissociation between implicit and explicit memory at encoding has been provided by two fMRI studies. One study was conducted by Schott et al. (2006) using the same paradigm as in the previous ERP experiment (Schott et al. 2002). They found that the recognition DM included enhanced activity in bilateral hippocampus and parahippocampal gyrus and left prefrontal cortex. In contrast, the priming-without-recognition DM included reduced activity in bilateral occipital and prefrontal cortex and left fusiform gyrus. Wimber et al. (2010) adopted a combined incidental perceptual identification and intentional recognition memory test to sort study trials into later remembered items (words that were later identified and correctly recognized as studied), later-primed items (words that were later identified but not recognized as studied), and later-nonidentified items (words that later elicited no identification response, or were incorrectly identified as other words). They found two distinct frontoparietal cortical networks that predicted later incidental perceptual identification priming and later intentional recognition memory. Activation in ventrolateral prefrontal areas and dorsal

posterior parietal regions predicted later recognition memory, whereas activation in dorsolateral and medial frontal areas and ventral posterior parietal cortex predicted later perceptual identification priming. Though previous studies have all emphasized the dissociation between implicit memory and explicit memory at encoding, the pattern of results across studies leaves several ambiguities. For example, some DM effects are shared by implicit and explicit memory (Meng, 2012); and the activation of specific brain regions (e.g., MTL or hippocampus) has predicted later perceptual priming in one study (Wimber et al., 2010), but was associated with later conscious recollection in another study (Schott et al., 2006). Wimber et al. (2010) attributed this contradiction with Schott et al. (2002) to the contamination of priming by familiarity. Familiarity, a form of explicit memory, is a fast, automatic process in which memory judgments can be driven by the increased fluency of reprocessing studied information (Taylor and Henson, 2012; Hayes and Verfaellie, 2012; Lucas et al., 2012). According to dual-process models of recognition memory, there are two processes contributing to recognition: recollection and familiarity (Yonelinas, 2002; Rugg and Curran, 2007). Recollection refers to the retrieval of contextual aspects of the encoding episode (normally leading to a “remember” response), whereas familiarity refers to a conscious impression that an item has been experienced recently, in the absence of any contextual associations (normally leading to a “know” response). According to Wimber et al. (2010), participants were likely to answer “studied” only if they specifically recollected the study episode or they were very sure that an item was studied, such that the primed item category might be subject to contamination by familiarity or by lowconfidence recognition memory. It is an open question as to whether the source of this increased fluency and its potential effect on priming and familiarity is the same or not (for a review, see Paller et al., 2007). It would be informative to determine if the neural events at encoding that predict later priming are distinct from those that predict later familiarity. On the other hand, though most studies are interested in the dissociation between recollection and familiarity, it is yet to be determined whether recollection and familiarity are supported by categorically different cognitive processes (Yonelinas, 2002; Woodruff et al., 2006), or whether they simply differ in terms of trace strength (Squire et al., 2007; Mickes et al., 2009). Several studies (Smith, 1993; Friedman and Trott, 2000; Mangels et al., 2001; Duarte et al., 2004; Voss and Paller, 2009b) have attempted to categorize study ERP findings on subsequent memory performance derived from remember/know recognition tests, but failed to find the difference between remember and know response (e.g. Friedman and Trott, 2000; Mangels et al., 2001). Mangels et al. (2001) attributed this failure to the mixture between K responses and “lucky guess” in general remember/know judgments. That is, participants might categorize an item as K on the basis of guessing which might driven by perceptual fluency. In the present study, we modified Meng's (2012) paradigm, including a “remember/know/guess” judgment after the twoalternative forced-choice recognition test, so as to concurrently acquire neural correlates of encoding processes

75

brain research 1585 (2014) 72–82

yielding recollection, familiarity and priming in a single test. Similar with Meng's (2012) analysis strategy, study trials were sorted as a function of the forced-choice recognition response and the subsequent judgment on the basis of the recognition response. A summary for the memory categories is illustrated in Fig. 1. Subsequent remembered trials were those studied items that were later recognized in the forced-choice recognition and indicated as being based on remembering, reflecting recollection of details from the study phase. Subsequent known trials were those studied items that were later recognized and indicated as being based on knowing, reflecting familiarity with no recollection of details from the study phase. Subsequent primed trials were those studied items that were later recognized but indicated as being based on guessing, reflecting priming driven by perceptual fluency from the study phase, in the absence of any concomitant awareness of memory retrieval. Subsequent forgotten trials were those studied items that were later not recognized in the forced-choice recognition test, that is, the new items paired with the studied items were incorrectly selected as studied, which reflected the lack of any memory for those studied items. The DM effects for recollection (DMR), familiarity (DMF) and priming (DMP) were evaluated separately by contrasting study ERPs corresponding to subsequent remembered, subsequent known or subsequent primed trials with study ERPs corresponding to subsequent forgotten trials. According to the Encoding Specificity Principle (Tulving and Thomson, 1973), specific encoding operations performed on what is perceived determine what is stored, and what is stored determines what retrieval cues are effective in providing access to what is stored. Therefore, we hypothesized that we would find a set of different components in the ERPs

Fig. 1 – Summary of the data categories.

recorded at encoding that would be sensitive to later recollection, familiarity and priming.

2.

Results

2.1.

Behavioral results

The percentage of responses to each condition and associated RTs in the test are shown in Table 1. For false alarm, there were too few trials to estimate RTs reliably. Hence, the result of false alarm did not appear in Table 1. There was a greater mean percentage of accurate recognition responses for remember [Rhit vs RFA: t(14) ¼10.214, po0.001] and Know [Khit vs KFA: t(14) ¼7.109, po0.001] judgments, but not for guess judgments [Ghit vs GFA: t(14)¼ 0.406, p¼ 0.691], indicating that conscious recognition can help participants respond more accurately, regardless of whether or not they retrieved details from the study phase. Consistent with this effect, we also found that, for hit responses in forced-choice recognition, the mean RTs increased monotonically with increasing recognition confidence [F(1.15,16.10)¼ 17.802, po0.001]. Further comparisons showed recognition responses for remember trials were faster than for know trials [p ¼0.002], and recognition responses for know trials were faster than for guess trials [p ¼0.010]. Another analysis of RTs for old and new recognition within Guess trials did not show a significant effect [Ghit vs GFA: t(14) ¼0.572, p¼ 0.577]. Taken together, we found that concomitant awareness of memory retrieval contributed to recognition performance, as reflected by faster, more accurate responses for R and K responses. We also found that R responses were made more rapidly than K responses, which was similar with previous findings (Dewhurst et al., 2006). Dewhurst et al. (2006) proposed that the faster RTs for R responses reflect the greater ease of making such decisions by mentally reinstating aspects of the encoding context. We failed to find evidence that perceptual fluency contributed to forced-choice recognition responding in the guess condition. Given that prior studies using a similar forced-choice recognition test found that responses in the forced-choice test were often accurate when explicit memory was introspectively absent (e.g., for guess trials), the failure in present experiment to find this effect may be attributable to insufficient data for a reliable

Table 1 – Mean percentages and mean RT (in ms) of responses for each condition (old/new) with each confidence category at test, with standard errors of the mean. Forced-choice response

Recognition confidence Remember

Know

Guess

Old (hit) % RT

25.0 (7) 979 (39)

21.5 (5) 1061 (51)

21.9 (5) 1131 (65)

New (false alarm) % RT

2.9 (5) –

7.7 (3) –

21.3 (6) 1124 (67)

SEM values are presented in parentheses; ––¼ insufficient numbers; RT is to the first forced-choice decision.

76

brain research 1585 (2014) 72–82

effect. An alternative explanation is that perceptual fluency is not enough to change behavioral responses, as behavioral data are not sensitive enough to demonstrate the presence of a priming effect under some conditions (Hertz et al., 1994; Begleiter et al., 1995).

2.2.

ERPs results

ERP waveforms elicited during study trials displayed P1 (about 104 ms), N1(about 146 ms) and P2 (about 208 ms) components at parietal–occipital sites (PO7/PO8), followed by a widespread negative deflection between 300 and 500 ms most prominent at anterior sites, and a long lasting late positive component (LPC) with a large amplitude was observed at posterior sites. Mean ERP amplitudes to four categories during study phase were computed at five separate sets of electrodes for prefrontal, frontal, central, parietal and occipital locations (as depicted in Figs. 2, 3, and 5) and analyzed over successive 100 ms time windows starting at 200 ms. According to our interest, we analyzed each DM effect for recollection(DMR), familiarity(DMF) and priming(DMP) by contrasting study ERPs of subsequent remembered, subsequent known or subsequent primed trials versus study ERPs of subsequent forgotten trials with DM (subsequent remembered/known/primed vs forgotten)  hemisphere ANOVAs for each electrode set.

2.2.1.

DMR effect

Fig. 2 shows the ERPs difference between remembered and forgotten study trials. As revealed by ANOVA, there were marginal interactions between DM and hemisphere from 300 ms, but the interactions were significant only from 900 ms to 1200 ms over prefrontal sites [900–1000 ms: F(1,14)¼ 4.672, p¼ 0.018; 1000–1100 ms: F(1,14)¼5.605, p¼0.019; 1100–1200 ms: F(1,14)¼ 5.595, p¼ 0.009]. Simple effect comparisons revealed the significant DM effects mainly at right prefrontal electrode from 300 ms to 500 ms and 900 ms to 1200 ms (all pso0.050), with a pattern of remembered being more positive-going than forgotten. The effect reversed in polarity over occipital electrodes, with a pattern of remembered being more negative-going than forgotten. This main DM effect was significant from 800 ms to the end of recording [800–900 ms: F(1,14)¼ 7.146, p¼0.018; 900–1000 ms: F(1,14)¼9.685, p¼0.008; 1000–1100 ms: F(1,14)¼9.185, p¼ 0.009; 1100–1200 ms: F(1,14)¼ 5.462, p¼ 0.035] and was not significantly lateralized.

2.2.2.

DMF effect

Fig. 3 shows the ERPs difference between known and forgotten study trials. As revealed by ANOVA, the DMF effect was associated with a marginal main DM effect during 300– 400 ms over prefrontal electrodes [F(1,14)¼ 3.383, p ¼0.087] and a main DM effect during 500–600 ms over parietal

Fig. 2 – DMR effect indicated by ERPs difference between subsequent remembered and subsequent forgotten study trials.

brain research 1585 (2014) 72–82

77

Fig. 3 – DMF effect indicated by ERPs difference between subsequent known and subsequent forgotten study trials.

Fig. 4 – Topographic maps depicting the DM effects of subsequent recollection and subsequent familiarity in 300–400 ms and 500–600 ms time windows. Small circles represent electrode locations as viewed from above.

electrodes [F(1,14)¼ 5.322, p ¼0.037], all with a pattern of known being more positive-going than forgotten. Topographical analyses were also made for DMR and DMF contrasts using the vector normalization approach (McCarthy and Wood, 1985; Picton and et al., 2000), in order to determine whether DMR and DMF reflected the engagement of distinct configurations of neural generators. The first comparison sought to directly compare DMR and DMF topographies for 300–400 ms latency interval in which both effects were

significant. A nonsignificant condition-by-electrode interaction [F(3.19,44.72)¼ 0.827, p¼ 0.493] indicated that similar prefrontal-DM topographies for recollection and familiarity. The second comparison sought to assess the consistency of the effects for each condition over time. Thus, the normalized difference topography for known minus forgotten was compared for the 300–400 ms and 500–600 ms latency intervals, and the same was done for remembered minus forgotten normalized difference topography. For DMF, a significant

78

brain research 1585 (2014) 72–82

Fig. 5 – DMP effect indicated by ERPs difference between subsequent primed and subsequent forgotten study trials.

interval-by-electrode interaction [F(4.11,57.61)¼ 5.819, po0.001] substantiated that the intervals (300–400 and 500–600 ms) captured contributions from distinct neural populations to DMF effect. A nonsignificant interaction was identified for DMR [F(4.25,59.52)¼0.385, p¼0.830], which indicated that DMR did not vary spatially across the two latency intervals. Taken together, these comparisons indicated that prefrontalpositive- going- DM during 300–400 ms was similar for recollection and familiarity. In addition, familiarity was associated with a positive-going transient DM effect at parietal sites in 500–600 ms, and recollection was associated with a later (beginning at 900 ms) right-prefrontal positive-going DM effect, and a later (beginning at 800 ms) occipital, though negativegoing DM effect (see Fig. 4).

2.2.3.

DMP effect

Fig. 5 shows the ERPs difference between primed and forgotten study trials. Qualitatively, the DMP appeared different from DMR and DMF over the anterior sites and similar over the posterior sites, all with the same pattern of primed being more negativegoing than forgotten. The ANOVA showed there was a marginal main DM effect from 700 ms until the end of recording (pso0.090) over prefrontal electrodes, significant only in 1000– 1100 ms [F(1,14)¼ 5.417, p¼ 0.035] and 1100–1200 ms [F(1,14)¼

4.914, p¼ 0.044] time windows. At frontal and central electrodes, the DMP onset at 600 ms, continuing until the end of recording (pso0.060), with the maximum during 700–800 ms [frontal: F(1,14)¼ 7.003, p¼ 0.019; central: F(1,14)¼7.548, p¼0.016]. Topographical analyses were also made for DMR and DMP contrasts since both effects were significant from 700 ms after stimulus onset. For each latency interval, DMR topography was compared to DMP topography and significant condition-byelectrode interaction was identified [700–800 ms: F(4.83,67.69)¼ 6.014, po0.001; 800–900 ms: F(5.99,83.93)¼ 2.307, p¼0.041; 900– 1000 ms: F(7.47,104.53)¼ 2.157, p¼ 0.040; 1000–1100 ms: F (5.39,75.44)¼ 3.143, p¼0.011; 1100–1200 ms: F(6.39,89.44)¼2.762, p¼0.015]. Based on these ERP findings, it appears that different DM effects recorded during encoding predicted subsequent recollection and priming. The recollection DM showed a more posterior distribution, while the priming DM showed a more anterior distribution (see Fig. 6).

3.

Discussion

By combining the DM paradigm and a forced-choice recognition test, we were able to directly compare neural correlates of encoding yielding subsequent recollection, familiarity, and

brain research 1585 (2014) 72–82

79

priming. The study trials were divided into four categories: subsequent remembered, subsequent known, subsequent primed and subsequent forgotten. Through comparing each of these conditions with forgotten trials separately, we observed different DM effects recorded during encoding predicted subsequent recollection, familiarity and priming.

3.1. Distinct correlates for recollection and familiarity at encoding The DM effect predicting later recollection was associated with a robust sustained (beginning at 300 ms) prefrontal positive-going DM which was right-lateralized, and a later (beginning at 800 ms) occipital, negative-going DM. In contrast, the DM effect predicting later familiarity was associated with two transient positive-going effects: an earlier (300– 400 ms) prefrontally distributed difference and a later (500– 600 ms) parietally distributed difference. As we mentioned before, several studies have compared DM effects associated with familiarity and recollection by remember/know recognition test (Smith, 1993; Friedman and Trott, 2000; Mangels et al., 2001; Duarte et al., 2004; Voss and Paller, 2009b), but failed to find differences, likely due to the mixture between K responses and “lucky guess” responses in a general remember/know recognition test. In the present study, we adopted a remember/know/guess judgment to distinguish responses based on knowing from responses based on guessing and found a difference between these two categories, discussed below. More importantly, we observed different neural events during encoding that predicted later familiarity versus later recollection. Familiarity was associated with an earlier (300–400 ms) prefrontal positive-going DM effect and a later (500–600 ms) parietal positive-going DM effect. These two positive-going DM effects are similar to those observed in previous studies, and have been attributed to neural processing in the prefrontal cortex (PFC) and medial temporal lobes (MTL) respectively (Donaldson and Rugg, 1999; Buckner et al., 2000; Ranganath et al., 2004; Staresina and Davachi, 2006; Summerfield et al., 2006). Evidence from neuroimaging studies suggests that the PFC–MTL alliance mediates the encoding of new information into explicit memory: PFC is assumed to process and organize incoming information (elaborative processing), which is later stored in the MTL to produce the memory trace for later explicit retrieval (Moscovitch, 1992; Simons and Spiers, 2003; Cabeza and Moscovitch, 2013). Perhaps consistent with these putative PFC–MTL interactions at encoding, the topography analysis of the familiarity DM effect in the present study showed a shift in distribution from prefrontal to parietal over time. Together, these results suggest that encoding processes supporting later familiarity-based responses may include not only perceptual or conceptual processing of study items, but also the operation of an explicit memory system. Notably, we observed a similar early prefrontal DM effect associated with subsequent recollection. We also observed a weak DM effect associated with recollection at parietal sites in the 500–600 ms time window (see Fig. 4), though pairwise comparisons failed to reveal any significant DM effect when comparing the ERPs of remembered with forgotten items. We suggest that this weak parietal DM effect associated with

Fig. 6 – Topographic maps depicting the DM effect of subsequent recollection and subsequent priming in 700– 1200 ms time windows. Small circles represent electrode locations as viewed from above.

recollection in the present study may be canceled out by later parietal–occipital effect, which was inverse but stronger. More importantly, a later but sustained (beginning at 800 ms) negativegoing DM effect over occipital sites was observed to be only associated with recollection. Similarly, Mangels et al. (2001) found subsequently remembered words elicited a sustained inferior temporal negativity (500–1500 ms) relative to subsequently known and missed items, whereas subsequently known words elicited only a transient left fronto-temporal negative wave (N340) relative to missed items. Taken together, these results might indicate that PFC–MTL interactions could support later “know” responses. Such processing might be necessary, but not sufficient, for later “remember” responses, which may require further retrieval of specific contextual detail from a learning episode. The posterior negative sustained potential found in the current study might indicate sustained activation of object representations pertaining to the concrete object the word represented or representations of the word itself (Mangels et al., 2001). These data collectively support the assertion that formation of vivid, “recollective” memory is thought to require more processes than formation of less detailed traces, as indexed by sustained occipital negative activity.

3.2. Correlates for priming at encoding and its dissociation from explicit memory Through a two-stage forced-choice recognition test including a guess response option, we obtained a purer measure of priming—one that is both involuntary and unassociated with awareness of prior study. This priming-DM effect was associated with a negative-going effect which onset at 600 ms, mainly distributed over anterior scalp sites. Though the negative-going DM effect associated with later priming was similar with those priming DM effects found in previous studies (Schott et al., 2002; Meng, 2012; 2013), the timing and topography were somewhat different. For example, in Schott et al. (2002) study, when compared to forgotten test trials, primed test trials were associated with more negative waveforms at study. This DM effect was observed over central and parietal electrodes starting as early as 200 ms after word onset. Meng observed a negative-going priming DM effect with a temporalmaximum beginning 200 ms using word stimuli (Meng, 2012),

80

brain research 1585 (2014) 72–82

and a 400–500 ms frontal–central negative-going DM effect with face stimuli (Meng, 2013). Consistent with previous studies, the timing and topography of DM effects varied depending upon the precise experimental conditions (for a review, see Qin et al., 2007), and thus differences in DM effects likely reflect differences in experimental design and analysis methods across studies. In summary, we combined a forced-choice recognition test with remember/know/guess judgments to differentiate recollection, familiarity and priming concurrently in one test, and observed clearly a neural distinction at encoding between priming and familiarity. Compared to subsequent forgotten trials, subsequent known trials were associated with a positive-going DM effect over prefrontal and parietal sites, while later primed trials were mainly associated with anterior negative-going DM effects. Therefore, the neural events at encoding that predict later priming were distinct from those that predict later familiarity. Furthermore, the priming-DM effect was also distinct from the recollection-DM effect, which appeared largely as a positive-going DM effect at prefrontal sites. By contrast, priming-DM effects manifested as a negative-going DM effect at the same sites and in the same time windows. This is generally consistent with fMRI studies which have found that explicit memory and priming are associated with distinct patterns of increased and decreased activity respectively in prefrontal cortex during both encoding (Schott et al., 2006) and retrieval (Donaldson et al., 2001). Although recollection was also associated with a later but sustained (beginning at 800 ms) negative-going DM effect over occipital sites in the present experiment, topographical differences between these effects were reliable after rescaling. This confirmed that recollection and priming exhibited different neural correlates at encoding. Collectively, these results raised the possibility that the anterior negative-going DM effect in the current study might be a “pure” correlate of encoding into the perceptual representation system, distinguishable from correlates of encoding into the semantic and episodic memory systems. Schott et al. (2002) proposed that modulations of this negative-going waveform associated with later priming may be related to the efficiency with which words are perceptually and lexically processed. Wimber et al. (2010) suggested that the negative DM effect might reflect stimulus-driven attentional orienting at encoding, which was unfavorable for later recognition memory, but facilitated later perceptual identification. The present findings support these accounts. Anterior negativegoing DM effect might index stimulus-driven perceptual processing. During encoding, if an item was just processed by such perceptual processing, it could not be successfully retrieved with awareness. But it might be retrieved in the absence of awareness in forced-choice recognition test, because its perceptual fluency was systematically greater than that of the accompanying new item. In conclusion, results from the current study provide evidence for distinct neural signals associated with recollection, familiarity and priming during encoding. These neural signals do not simply differ in terms of trace strength, instead, they represent different cognitive processes underlying the encoding of verbal information into episodic memory, which separately supported later remembering, knowing and priming.

4.

Experimental procedure

4.1.

Participants

Eighteen undergraduate students (ten women; mean age, 21.8 years old; range, 20–24 years old) at Fujian Normal University gave written consent and were paid to participate in the study. All participants reported themselves to be in good health and normal or corrected-to-normal vision, with no history of neurological illness. Three participants' ERP data were excluded from the analysis because they fail to get more than 20 artifact free trials in any condition of interest.

4.2.

Stimuli

480 two-character Chinese emotionless words of low frequency were selected from the Modern Chinese Word Frequency Dictionary (1986) compiled by the Institute of Language Teaching and Research. The frequencies of words varied from 2.3 to 9.9 occurrences per million, with a mean frequency 3.7 occurrences per million. The words were divided randomly into four sets of 120 each, such that the frequency, stroke, pronunciation and configuration of words for the sets were matched. Two sets were assigned to the study condition, and the other two sets were presented at the forced-choice recognition test as new items paired with the studied items.

4.3.

Procedure

The participants sat on a sofa in a sound insulated, electromagnetically shielded room and held a game control handle. The experimental procedure was programmed in Presentation 0.71, and run on a DELL Dimension 8200 computer. All stimuli were presented centrally on the screen of a monitor placed approximately 80 cm from participants. The participants were tested in two study-test blocks and given short breaks (5710 min) between each study-test block. Each block consisted of three parts: a study phase, a filled delay, and a test phase (see Fig. 7). Practice was conducted before the experiment. During the study phase, a white pane of 10.58  7.06 cm2 was displayed in the central visual field on the screen against black background, and 120 words were presented sequentially in the center of the pane, with half shown in red and half in blue. Participants were instructed to judge the color of each word with the index fingers (left button¼ blue; right button ¼red) and the response hand was counterbalanced

Fig. 7 – Diagram of experiment procedure.

brain research 1585 (2014) 72–82

across participants. Words and corresponding color were presented randomly for each participant. Each study trial consisted of the presentation of a word for 800 ms, a further fixation ranging 1400–1600 ms (Inter-stimulus interval, ISI). Each study block also included two filler items at the beginning and in the end which were not included in the recognition test, in order to reduce primacy and recency effects (Voss et al., 2008). Before each study phase, participants were instructed to not to try to memorize the words because doing so would interfere with the task at hand. They were also instructed to avoid blinking. During the filled delay, a 3-digit number was presented, and subjects were asked to subtract by threes and give verbal responses for a total of 1 min. There were two stages during the forced-choice recognition test phase. In the first stage, two words (one from study list, and one new) appeared concurrently in the center of a white pane of 19.05  7.06 cm2 on the black-background screen. The target (studied) word was equally likely to appear on the left or right side, and participants were instructed to indicate the side of the target by pressing button with the left or right index finger via the game control handle. Before each study phase, participants were informed that each test trial had a studied word, and if they could not recognize the target, a guess was permitted to help choose one. Each old/new pair trial appeared for 2000 ms, and participants were told to respond as accurately and quickly as possible after the words appeared on the screen. In the second stage, a cue to make a confidence judgment appeared directly following the recognition response, and participants indicated whether the foregoing studied-selection was based on remembering, knowing or guessing. Remember indicated the retrieval of specific details from the study phase supporting the recognition decision. Know indicated recognition supported only by “a weak feeling of familiarity” with no details from the study phase retrieved. Guess indicated “absolutely no feeling of memory” such that the stimulus in no way felt “old”. Subjects were told to press this button when they were “just randomly guessing because a response is required for each trial.” The cue was presented for 1400 ms and composed of a display of the three response options. Responses were made via the game control handle, left button with left index finger for remember response, right upper button with right index finger for know response, and right bottom button with right thumb for guess response. All words in the test were in black.

4.4.

ERP recording and analysis

Electroencephalographic (EEG) recordings were recorded from 62 Ag/AgCl electrodes embedded in a cap of ESI-64-channel electrophysiology recording system produced by Neuroscan Co. The electrodes were located according to an extension of the international 10–20 system. Vertical electrooculogram (EOG) was recorded by electrodes placed directly above and below the supra-orbital of left eye, and horizontal EOG and blinks were monitored via a bipolar montage at the outer canthi of each eye. Online reference electrode was on Vertex. The ground electrode was between Fpz and Fz. The impedances between scalp and electrodes were kept below 5 KΩ. EEG signals were filtered with a band-pass of 0.05–100 Hz and sampled at a rate of 500 Hz.

81

Study EEG recordings were re-referenced offline to the average of the left and right mastoid recordings and were filtered with a band-pass of 0.05–100 Hz and sampled at a rate of 500 Hz. ERPs were obtained by averaging EEG recordings, time-locked to the onset of the study word, over epochs of 1100 ms, including a 200 ms baseline prior to word onset. Only those study trials that fell into the subsequent remembered, subsequent known, subsequent primed, and subsequent forgotten categories were used in averaging. Among these trials, any exceeding 775 μV at any electrode was excluded from analysis, as were trials with artifacts in the EOG channels. All conditions included in statistical analyses were required to have a minimum of 20 artifact free trials per participant in each condition. Statistical analysis of the study ERPs involved repeated measures ANOVA (α¼ 0.05) on mean amplitudes in every 100 ms time windows from 300 ms post stimulus onset, and for five separate sets of electrodes (the prefrontal electrodes FP1/FPz/FP2; the frontal electrodes F3/Fz/F4; the central electrodes C3/Cz/C4; the parietal electrodes P3/Pz/P4; and the occipital electrodes O1/Oz/O2). The data analysis was performed by SPSS 10.0 software and supplemented with pairwise comparisons or simple effect comparisons when appropriate. The Greenhouse–Geisser correction for nonsphericity was used when necessary and Bonferroni correction was used for multiple comparisons.

r e f e r e nc e s

Begleiter, H., Porjesz, B., Wang, W., 1995. Event-related brain potentials differentiate priming and recognition to familiar and unfamiliar faces. Electroencephalogr. Clin. Neurophysiol. 94, 41–49. Buckner, R.L., Koutstaal, W., Schacter, A.L., Rosen, B.R., 2000. Functional MRI evidence for a role of frontal and inferior temporal cortex in amodal components of priming. Brain 123, 620–640. Cabeza, R., Moscovitch, M., 2013. Memory systems, processing modes, and components: functional neuroimaging evidence. Perspect. Psychol. Sci. 8 (1), 49–55. Dew, I.T.Z., Cabeza, R., 2011. The porous boundaries between explicit and implicit memory: behavioral and neural evidence. Ann. N. Y. Acad. Sci. 1224, 174–190. Dewhurst, S.A., Holmes, S.J., Brandt, K.R., 2006. Measuring the speed of the conscious components of recognition memory: remembering is faster than knowing. Conscious. Cogn. 15 (1), 147–162. Donaldson, D.I., Petersen, S.E., Buckner, R.L., 2001. Dissociating memory retrieval processes using fmri: evidence that priming does not support recognition memory. Neuron 31 (6), 1047–1059. Donaldson, D.I., Rugg, M.D., 1999. Event-related potential studies of associative recognition and recall: electrophysiological evidence for context dependent retrieval processes. Cogn. Brain Res. 8, 1–16. Duarte, A., Ranganath, C., Winward, L., Hayward, D., Knight, R.T., 2004. Dissociable neural correlates in familiarity and recollection during the encoding and retrieval of pictures. Cogn. Brain Res. 18, 255–272. Friedman, D., Ritter, W., Snodgrass, J.G., 1996. ERPs during study as a function of subsequent direct and indirect memory testing in young and old adults. Cogn. Brain Res. 4, 1–13.

82

brain research 1585 (2014) 72–82

Friedman, D., Trott, C., 2000. An event-related potential study of encoding in young and older adults. Neuropsychologia 38, 542–557. Guo, C.Y., Voss, J.L., Paller, K.A., 2005. Electrophysiological correlates of forming memories for faces, names, and facename associations. Cogn. Brain Res. 22, 153–164. Hayes, S.M., Verfaellie, M., 2012. The impact of fluency on explicit memory tasks in amnesia. Cogn. Neurosci. 3 (3–4), 216–217. Henson, R.N., 2003. Neuroimaging studies of priming. Prog. Neurobiol. 70, 53–81. Henson, R.N., Goshen-Gottstein, Y., Ganel, T., Otten, L.J., Quayle, A., Rugg, M.D., 2003. Electrophysiological and haemodynamic correlates of face perception, recognition and priming. Cereb. Cortex 13, 793–805. Hertz, S., Porjesz, B., Begleiter, H., Chorlian, D., 1994. Event-related potentials to faces: the effects of priming and recognition. Electroencephalogr. Clin. Neurophysiol. 92, 342–351. Lucas, H.D., Paller, K.A., Voss, J.L., 2012. On the pervasive influences of implicit memory. Cogn. Neurosci. 3 (3-4), 219–226. Mangels, J.A., Picton, T.W., Craik, F.I.M., 2001. Attention and successful episodic encoding: an event-related potential study. Cogn. Brain Res. 11, 77–95. McCarthy, G., Wood, C.C., 1985. Scalp distributions of eventrelated potentials: an ambiguity associated with analysis of variance models. Electroencephalogr. Clin. Neurophysiol. 62, 203–208. Meng, Y.F., Guo, C.Y., 2006. Dissociations between implicit and explicit memory: an ERP study of face recognition. Acta Psychol. Sin. 38 (1), 15–21 (in Chinese). Meng, Y.F., 2012. ERP dissociation and connection between implicit and explicit memory at encoding. Acta Psychol. Sin. 44 (1), 30–39 (in Chinese). Meng, Y.F., 2013. Neural correlates at encoding of covert face recognition. Acta Psychol. Sin. 45 (1), 30–39 (in Chinese). Mickes, L., Wais, P.E., Wixted, J.T., 2009. Recollection is a continuous process: implications for dual-process theories of recognition memory. Psychol. Sci. 20 (4), 509–515. Moscovitch, M., 1992. Memory and working-with-memory: a component process model based on modules and central systems. J. Cogn. Neurosci. 4, 257–267. Paller, K.A., Voss, J.L., Boehm, S.G., 2007. Validating neural correlates of familiarity. Trends Cogn. Sci. 11, 243–250. Paller, K.A., Hutson, C.A., Miller, B.B., Boehm, S.G., 2003. Neural manifestations of memory with and without awareness. Neuron 38, 507–516. Paller, K.A., 1990. Recall and stem-completion priming have different electrophysiological correlates and are modified differentially by directed forgetting. J. Exp. Psychol.: Learn. Mem. Cogn. 16 (6), 1021–1032. Paller, K.A., Kutas, M., Mayes, A.R., 1987. Neural correlates of encoding in an incidental learning paradigm. Electroencephalogr. Clin. Neurophysiol. 67, 360–371. Picton, T.W., Bentin, S., Berg, P., Donchin, E., Hillyard, S.A., Johnson, R.J.R., et al., 2000. Guidelines for using human eventrelated potentials to study cognition: recording standards and publication criteria. Psychophysiology 37, 127–152. Qin, S.Z., Han, B.X., Luo, J., 2007. The neural mechanisms underlying successfully encoding episodic memories: evidence from the subsequent memory paradigm. Adv. Psychol. Sci. 15 (3), 401–408. Ranganath, C., Yonelinas, A.P., Cohen, M.X., Dy, C.J., Tom, S.M., D’Esposito, M., 2004. Dissociable correlates of recollection and familiarity within the medial temporal lobes. Neuropsychologia 42, 2–13.

Rugg, M.D., Curran, T., 2007. Event-related potentials and recognition memory. Trends Cogn. Sci. 11 (6), 251–257. Rugg, M.D., Mark, R.E., Walla, P., Schloerscheidt, A.M., Birch, C.S., Allan, K., 1998. Dissociation of the neural correlates of implicit and explicit memory. Nature 392, 595–598. Schott, B.H., Richardson-Klavehn, A., Heinze, H.J., Du¨zel, E., 2002. Perceptual priming versus explicit memory: dissociable neural correlates at encoding. J. Cogn. Neurosci. 14 (4), 578–592. Schott, B., Henson, R.N., Richardson-Klavehn, A., Becker, C., Thoma, V., Heinze, H.J., et al., 2005. Redefining implicit and explicit memory: the functional neuroanatomy of priming, remembering, and control of retrieval. Proc. Natl. Acad. Sci. USA 102, 1257–1262. Schott, B.H., Richardson-Klavehn, A., Henson, R.N., Becker, C., Heinze, H.J., Du¨zel, E., 2006. Neuroanatomical dissociation of encoding processes related to priming and explicit memory. J. Neurosci. 26 (3), 792–800. Simons, J.S., Spiers, H.J., 2003. Prefrontal and medial temporal lobe interactions in long-term memory. Nat. Rev. Neurosci. 4, 637–648. Smith, M.E., 1993. Neurophysiological manifestations of recollective experience during recognition memory judgments. J. Cogn. Neurosci. 5, 1–13. Sanquist, T.F., Rohrbaugh, J.W., Syndulko, K., Lindsley, D.B., 1980. Electrocortical signs of levels of processing: perceptual analysis and recognition memory. Psychophysiology 17, 568–576. Squire, L.R., Wixted, J.T., Clark, R.E., 2007. Recognition memory and the medial temporal lobe: a new perspective. Nat. Rev. Neurosci. 8, 872–883. Staresina, B.P., Davachi, L., 2006. Differential encoding mechanisms for subsequent associative recognition and free recall. J. Neurosci. 26, 9162–9172. Summerfield, C., Greene, M., Wager, T., Egner, T., Hirsch, J., Mangels, J., 2006. Neocortical connectivity during episodic memory formation. PLoS Biol. 4, e128, http://dx.doi.org/ 10.1371/journal.pbio.0040128. Taylor, J.R., Henson, R.N., 2012. You can feel it all over: many signals potentially contribute to feelings of familiarity. Cogn. Neurosci. 3 (3-4), 209–210. Tulving, E., Thomson, D.M., 1973. Encoding specificity and retrieval processes in episodic memory. Psychol. Rev. 80, 352–373. Voss, J.L., Paller, K.A., 2008. Brain substrates of implicit and explicit memory: the importance of concurrently acquired neural signals of both memory types. Neuropsychologia 46, 3021–3029. Voss, J.L., Baym, C.L., Paller, K.A., 2008. Accurate forced-choice recognition without awareness of memory retrieval. Learn. Mem. 15, 454–459. Voss, J.L., Paller, K.A., 2009a. An electrophysiological signature of unconscious recognition memory. Nat. Neurosci. 12, 349–355. Voss, J.L., Paller, K.A., 2009b. Remembering and knowing: electrophysiological distinctions at encoding but not retrieval. NeuroImage 46, 280–289. Wimber, M., Heinze, H.J., Richardson-Klavehn, A., 2010. Distinct frontoparietal networks set the stage for later perceptual identification priming and episodic recognition memory. J. Neurosci. 30 (40), 13272–13280. Woodruff, C.C., Hayama, H.R., Rugg, M.D., 2006. Electrophysiological dissociation of the neural correlates of recollection and familiarity. Brain Res. 1100, 125–135. Yonelinas, A.P., 2002. The nature of recollection and familiarity: a review of 30 years of research. J. Mem. Lang. 46, 441–517.