Research ReportSource recognition by stimulus content in the MTL
Introduction
Recognition of a previously encountered event can be based on remembering the event with its contextual details (recollection) or an acontextual sense of prior occurrence of the event (familiarity) (Mandler, 1980, Yonelinas, 2002). Medial temporal lobe (MTL) subregions differ in their contribution to recollection and familiarity. That is, the hippocampus makes critical contributions to recollection but not familiarity, and the parahippocampal cortex is particularly involved in recollection of source information. However, the perirhinal cortex is thought to be related to familiarity (Brown and Aggleton, 2001, Holdstock et al., 2005, Manns and Eichenbaum, 2006). MTL subregions also differ in the types of stimulus content that they process (Awipi and Davachi, 2008, Graham et al., 2010, Lee et al., 2008, Liang et al., 2013, Taylor et al., 2007, Watson et al., 2012). Given that associations mediated by the hippocampus have been found with different types of stimuli, the hippocampus is considered to play more of a general role in association of items and contexts across content (Cansino et al., 2002, Davachi et al., 2003, Duarte et al., 2011, Staresina et al., 2011). Other regions tend to take more specific roles in terms of stimulus content: the parahippocampal cortex is recruited for visuo-spatial processing, whereas the perirhinal cortex is linked to object processing (Awipi and Davachi, 2008, Ekstrom and Bookheimer, 2007, Staresina et al., 2011, Wolbers and Büchel, 2005). Based on the role of the perirhinal cortex in object processing, it has been proposed that source memory for objects can be mediated by perirhinal activity, especially when the context is processed as a feature of the object (Diana et al., 2007, Ranganath, 2010).
While previous fMRI studies have provided useful information about neural activity related to content-selective source recognition, they have mostly used semantic tasks for examining neural correlates of contextual associations during retrieval that depend on content (Duarte et al., 2011, Dulas and Duarte, 2012). A semantic task for the formation of contextual associations is more likely to demand semantic processing of the to-be-remembered material. Previous neuroimaging studies of memory have shown that neural correlates of memory dissociate semantic and non-semantic memory (Otten and Rugg, 2001, Park et al., 2008), and these findings suggest that source memory with a semantic task may recruit neural correlates that are different from the neural correlates of source memory with a non-semantic task. While the semantic task provides an important path for encoding and retrieval of contextual associations, fMRI examinations of content-selective source recognition with non-semantic tasks are lacking.
Further, the results on content-selective effects on source recognition were rather mixed (e.g., Duarte et al., 2011), although hippocampal activity was reported for source recognition common to different stimulus content. When an object-based perceptual encoding task was used, perirhinal activity was related to source memory for objects, while parahippocampal activity predicted source memory for scenes, thereby revealing content-selective source encoding effects in the MTL cortex between two types of pictorial content (Staresina et al., 2011). Yet, it has not been established if content-selective source effects appear during retrieval when contextual associations are formed via object-based encoding, especially between verbal and pictorial content. Previous EEG studies demonstrated that words and pictures showed different congruency effects or different frontal effects (Bauch and Otten, 2012, Galli and Otten, 2011), and subsequent associative memory effects also dissociated between word and picture pairs (Park and Rugg, 2011). These findings suggested that neural correlates of content-selective effects occurred mainly between verbal and pictorial content. Based on these findings, the present study examined content-selective effects in source recognition by employing words and pictures. Object processing with words may require neural correlates that are different from object processing with pictures, considering that the generation of object representations from words may require additional processing than the generation of object representations from pictures due to the difference in the sensory process or the process match effects (Kinjo and Snodgrass, 2000). Alternatively, it is also plausible that both words and pictures recruit common neural correlates for source recognition as they form an integrated object representation and retrieve the representation regardless of content. Thus, it is of interest whether source recognition of words and pictures via object-based processing would show content-selective effects of source retrieval.
The MTL including the hippocampus is known to be sensitive to detecting the presence of novel stimuli (O’Kane et al., 2005, Stark and Okado, 2003). However, it has not yet been fully understood whether MTL novelty effects differ according to the content of new stimuli that are presented with studied stimuli during source recognition. On item recognition tests, the new items that were correctly judged as “new” (correct rejections) tended to elicit greater MTL activity than did the studied items that were correctly judged as “old” (hits) (Daselaar et al., 2006, Dudukovic and Wagner, 2007, O’Kane et al., 2005, Yu et al., 2012). These novelty detection effects have been interpreted as an MTL-mediated encoding process for new stimuli at test (Rugg et al., 2012, Stark and Okado, 2003). The hippocampus is sensitive to detection of new stimuli among studied stimuli between study and test, and the adjacent MTL cortex also tends to be sensitive to the difference between first-presented items versus repeated items on continuous recognition (Howard et al., 2011, Yassa and Stark, 2008). In light of content-selective functions of MTL subregions, it would be of interest to see if novelty detection effects between study and test also reveal differences of MTL subregions depending on stimulus content.
The present event-related fMRI addressed the question of whether retrieval of source memory elicits neural activity selective to stimulus content within the MTL. To this end, we manipulated the type of stimulus content (verbal vs. pictorial) with perceptual context (4 colors) through integrated contextual associations. Participants were presented an object either as a word or a picture in one of four color squares during study. To facilitate the formation of item-context associations, participants were instructed to form an image of the object in the color. During test, participants were presented stimuli without color and were asked to make source recognition judgments. For studied stimuli, content type was kept constant between study and test.
We first examined if source retrieval effects invariant of content would emerge in the hippocampus through integrated contextual associations. We hypothesized that source retrieval effects invariant of content would emerge in the hippocampus, as hippocampal source recognition effects have been found across different materials, tasks, and contexts. Second, we investigated if content-selective source retrieval effects would occur when both words and pictures underwent the same study task. We suspected that a greater emphasis would be placed on processing the word as an object in the word condition than in the picture condition. We hypothesized that content-selective source recognition effects would be found in different MTL regions, as it was more likely that the word with accurate source information had been processed as an object than the word without source information due to the nature of the study task. That is, the degree of object processing would be more critical in source recognition of words than pictures, as pictures were already presented in a visual object form more readily available for association to context. In contrast, words would require additional processing to first generate object representations before engaging in association to context. This difference in object processing was expected to bestow words a memorial advantage over pictures in behavioral performance. Functionally, it was assumed that the difference between source correct words and source incorrect words would involve activity in the perirhinal cortex, based on previous findings of perirhinal activity in object processing (Bellgowan et al., 2009, Lee et al., 2006, Staresina et al., 2011, Watson et al., 2012). However, pictures with accurate source information and pictures without such information would not show differences in the perirhinal cortex as much as words would, considering that pictures were presented in visual form during study. Instead, more details of each picture, such as visuo-spatial details, would reflect differences between source correct pictures versus source incorrect pictures. Such differences would appear in the posterior MTL region, given that previous findings showed posterior MTL activity in memory for pictorial stimuli across different conditions (Gottlieb et al., 2012, Köhler et al., 2000, Park et al., 2012, Stark and Okado, 2003;). Third, we probed if neural activity sensitive to detection of new stimuli would show content-invariant and content-selective effects within the MTL. Based on previous findings of hippocampal novelty effects (Daselaar et al., 2006, Stark and Okado, 2003, Suzuki et al., 2011), we hypothesized that the anterior hippocampus would show content-invariant effects sensitive to detection of novel items, and the adjacent MTL regions would reveal content-selective novelty detection effects.
Section snippets
Behavioral results
The mean proportions of memory judgments and reaction times (RTs) for studied and new stimuli are displayed in Table 1. The adjusted estimate of source memory by discriminability, Pr (Psource hits–Psource false alarms) showed that accuracy of source judgments was greater for words (.56±24) than for pictures (.43±.26), t[23]=4.56, p <001, although source accuracy was significantly above the chance level of.25 for both words and pictures, t׳s[23]>3.32, p׳s<.005. The proportions of misses for
Discussion
The present study examined the question of whether contextual associations via an integrated perceptual task showed MTL source memory effects selective to stimulus content during retrieval. Source recognition effects common to both words and pictures were found in the left anterior hippocampus extending to the perirhinal cortex and the right posterior hippocampus. Word-selective source retrieval was related with activity in the left perirhinal cortex while picture-selective source recognition
Conclusions
The present study investigated source recognition that depends on stimulus content via integrated encoding within the MTL. The findings demonstrate that the hippocampus is pivotal in source recognition in both content-invariant and content-selective ways. Word-selective source recognition was associated with left perirhinal activity, reflecting the contribution of more extensive object processing for words with correct source information compared to words with incorrect source information. On
Participants
Twenty-four volunteers (16 female) between 18 and 29 years of age took part in the study. All participants were right-handed and native English speakers with no reported history of neurological or psychological disorders. Informed consent was obtained prior to participation under a protocol approved by the Institutional Review Boards of the University of Texas at Arlington and the University of Texas Southwestern Medical Center. Participants were compensated for their time.
Materials
A total of 300 object
References (57)
- et al.
Imaging recollection and familiarity in the medial temporal lobe: a three-component model
Trends Cognit. Sci.
(2007) - et al.
The effect of stimulus content on the neural correlates of source recollection
Brain Res.
(2011) - et al.
Classical and Bayesian inference in neuroimaging: application
Neuroimage
(2002) - et al.
Event-related fMRI: characterizing differential responses
Neuroimage
(1998) - et al.
Going beyond LTM in the MRL: a synthesis of neuropsychological and neuroimaging findings on the role of the medial temporal lobe in memory and perception
Neuropsychologia
(2010) - et al.
Episodic encoding and recognition of pictures and words: role of the human medial temporal lobes
Acta Psychol.
(2000) - et al.
Perirhinal cortex activity during visual object discrimination: an event-related fMRI study
Neuroimage
(2006) - et al.
Role of perirhinal cortex in object perception, memory and associations
Curr. Opin. Neurobiol.
(2001) - et al.
Neural activity supporting the formation of associative memory versus source memory
Brain Res.
(2012) - et al.
Item memory, context memory and the hippocampus: fMRI evidence
Neuropsychologia
(2012)