“Disorganized in time”: Impact of bottom-up and top-down negative emotion generation on memory formation among healthy and traumatized adolescents

Abstract

“Travelling in time,” a central feature of episodic memory is severely affected among individuals with Post Traumatic Stress Disorder (PTSD) with two opposite effects: vivid traumatic memories are unorganized in temporality (bottom-up processes), non-traumatic personal memories tend to lack spatio-temporal details and false recognitions occur more frequently that in the general population (top-down processes). To test the effect of these two types of processes (i.e. bottom-up and top-down) on emotional memory, we conducted two studies in healthy and traumatized adolescents, a period of life in which vulnerability to emotion is particularly high. Using negative and neutral images selected from the international affective picture system (IAPS), stimuli were divided into perceptual images (emotion generated by perceptual details) and conceptual images (emotion generated by the general meaning of the material). Both categories of stimuli were then used, along with neutral pictures, in a memory task with two phases (encoding and recognition). In both populations, we reported a differential effect of the emotional material on encoding and recognition. Negative perceptual scenes induced an attentional capture effect during encoding and enhanced the recollective distinctiveness. Conversely, the encoding of conceptual scenes was similar to neutral ones, but the conceptual relatedness induced false memories at retrieval. However, among individuals with PTSD, two subgroups of patients were identified. The first subgroup processed the scenes faster than controls, except for the perceptual scenes, and obtained similar performances to controls in the recognition task. The second subgroup group desmonstrated an attentional deficit in the encoding task with no benefit from the distinctiveness associated with negative perceptual scenes on memory performances. These findings provide a new perspective on how negative emotional information may have opposite influences on memory in normal and traumatized individuals. It also gives clues to understand how intrusive memories and overgeneralization takes place in PTSD.

Introduction

Episodic memory enables mental time travel in subjective time (Tulving, 2002, Tulving, 2005) and this “special, and unique relationship” of episodic memory to time, “surprisingly, is not widely known […] allowing one to re-experience, through autonoetic awareness, one’s own previous experiences” (Tulving, 1972). His statement has previously been expressed by Wells (1898/2002) who writes: “You are wrong to say that we cannot move about in Time. For instance, if I am recalling an incident very vividly I go back to the instant of its occurrence: I become absent-minded, as you say. I jump back for a moment.” Locke (1690/1975) has stated a long time ago that memory is the power “to revive perceptions, which it has once had, with this additional perception annexed to them, that it has had them before.” Episodic memory is pivotal in autobiographical memory formed of different types of representations, from general knowledge about oneself (semantic component) to very specific personal events (Conway, 2001). The way one remembers our past may be altered by self-conception (Wilson and Ross, 2003) and reciprocally. Hence impairment in episodic memory of personal events may disrupt the sense of continuity and coherence of the self (Conway and Pleydell-Pearce, 2000). Episodic memory can be severely affected in individuals with PTSD with two opposite effects. Traumatic memories arise suddenly without any effort of recollection and even against one’s effort to forget. These memories are unorganized in temporality and so vivid that the subject can feel he/she is actually reexperiencing the same event as the one that initially led to PTSD (bottom-up processes). Conversely, there is a trend to lose vividness and distinctiveness of other personal memories that become overgeneralized (top-down processes). Moreover, some clinical studies have shown that false memories (i.e., “memories” of personal facts that never happened) are more numerous in that population compared to healthy subjects (Brennen et al., 2007). Similarly, laboratory studies have shown that false recognitions are more frequent in PTSD compared to healthy subjects (Jelinek et al., 2009).

One of the main symptoms of PTSD is the intrusive recollection of highly specific single details of the traumatic event, criterion B of the definition of PTSD in DSM-IV-TR (American Psychiatric Association, 2000) with the patient acting or feeling as if the traumatic events were recurring. Temporality is severely altered in the sense that “what surges up in consciousness is not the memory of a “traumatic feeling” but a new feeling experienced in the present” (James, 1890). It initially takes the form of an unrelated set of sensory-perceptual details that only over time may come to be associated with a more abstract general event or lifetime period (Van der Kolk and Fisler, 1995, Howe and Courage, 1997). According to Janet (1894) “traumatic memory” repeats the past, while “narrative memory” narrates the past as past. More recently, Brewin and collaborators suggest that memories of a personally experienced traumatic event can be stored in two different representational formats (Brewin et al., 1996; Hellawell and Brewin, 2004). One format consists of a memory available for verbal communication, which can be retrieved either automatically or using deliberate and strategic processes, i.e. top-down processes. The second format includes trauma-related dreams and “flashbacks” and emanates from low-level perceptual processing of the traumatic scene via sensory information that receives little conscious processing, i.e. bottom-up processes. Visual imagery is predominant in traumatic memories, although other sensory-perceptual details may be the first trigger. They are commonly associated with intense personal distress. A patient has written that “fragments of horrific experiences are captured in the “brain’s eye”. … Those pictures … are frozen in time. On any given day … the pictures come to life again – the horror of what happened is once again very real, and it is happening all over again”. A young child with PTSD came to our consultation asking us “to cure her eyes … to wipe visions that was imposed to her”. A study using fMRI (Yang et al., 2004) indicated that patients with PTSD, compared to healthy subjects, both having experienced an earthquake, showed significantly greater activation in posterior visual (occipital) areas when evocating mental imagery about the trauma. Heightened proneness to form false memories in some specific laboratory tasks has also been shown (Jelinek et al., 2009) particularly with visual cues.

It is now well established that emotion influences memory formation (Buchanan, 2007). Most investigations have focused on how arousal (i.e., soothing or exciting effect of the stimulus) and valence (i.e., positive or negative) may interact with different processes, including attentional, perceptual, and conceptual processes (LaBar and Cabeza, 2006). Concerning explicit memory, there are two contrasting theories. The first proposes that emotion enhances recollective distinctiveness based on results showing that arousal increases memory vividness (Rimmele et al., 2011). Compared to words, perceptual stimuli, such as emotional pictures, seem more likely to elicit this recollection process. The second theory considers emotion as an organizational factor in memory that may enhance false memories related to the gist of the emotional target items. This conceptual relatedness hypothesis may be observed for both emotional words and pictures (Gallo et al., 2009). Although arousal is behind many significant changes in memory, from perception to consolidation (Mather and Sutherland, 2011), it has been clear for some time that the emotional valence of information also enhances memory, in a number of different ways. Behavioral studies have revealed that negative compared to neutral emotion, may enhance memory for specific details and leads to vivid recollection (Mickley and Kensinger, 2008). Nonetheless, negative emotional pictures may also increase confusability in memory and lead to false recollection (Gallo et al., 2009). This discrepancy suggests that some properties, specific to the pictures, may influence the recollection process.

More recently, two supplementary distinctions have been studied, perceptual versus conceptual processing, in addition to arousal and valence. These former types of processing refer to two distinct functional networks (Ritchey et al., 2011): processing perceptual emotional features relies particularly on an automatic pathway that includes the amygdale, while deep processing, focused on the meaning or interpretation of the emotional pictures, depends on prefrontal regions. These two distinct cortical networks overlap with those implicated in emotion generation: a bottom-up perception of aversive images and a top-down interpretation implicating conceptual understanding – including specific emotional state – of the emotional images (Ochsner et al., 2009).

Results on emotional memory in adolescence are in line with results in adult data showing a more pronounced memory bias for negative valence, for both hits (recognition) (D’Acremont and Van der Linden, 2007, Krauel et al., 2009) and false alarms (Nelson et al., 2003). Yet, neuroimaging studies have revealed that, unlike adults, adolescents are influenced more by the emotional nature of the stimuli than by the attentional demands of the task (Blakemore, 2008). Hare et al. (2008) have also reported that adolescents show exaggerated amygdala activity relative to children and adults during an emotional inhibition task. Wang et al. (2008) reported greater emotional interference in executive functioning in adolescents compared to adults when performing an emotional oddball paradigm featuring a human expression (sad or neutral). The immature top-down control and an impaired regulation of emotion is often mentioned in the literature (Galvan, 2010). This cognitive control of affective networks, enabling the coordination between the executive and emotional systems, undergoes gradual maturation during adolescence (Yurgelun-Todd, 2007). Hence, bottom-up perceptual and top-down conceptual processes implicated in emotion generation may differentially affect memory, and adolescence constitutes an appropriate model to examine this question.

Most theories about emotional memory are also valid for memory in PTSD. However, they may be inefficient to explain all specificities of emotional memory in that population. Studies about emotional memory in PTSD have tried to explain the origin of intrusive memories, whether in the form of “flashbacks” or not. Fewer studies have investigated specificities of episodic memory according to the arousal and valence of the material to remember, notwithstanding items were not directly linked to the trauma. Contrasting results may be explained notably by activation of different brain networks according to the prevalence of bottom-up or top-down processes.

Different experimental designs have been used to study these different types of emotional memory. Concerning intrusive memories, according to fear conditioning models, PTSD is characterized by an exaggerated response of the amygdala to threat-related stimuli (St Jacques et al., 2011), hence of prevalent bottom-up processes. Less is known about the phenomenon of overgeneralization and false memories, particularly for emotional memories, which we will refer to as “numbing in time.” Performance on declarative memory tests with neutral stimuli has been found to be consistently impaired in PTSD (Brewin et al., 2007; Johnsen and Asbjornsen, 2008), although results are far more complex for emotional memories. Most studies found, among patients with PTSD, a threat-related memory bias characterized by a retrieval “advantage” for stimuli related to the trauma or for other specific threats compared to non-traumatic stimuli (Brewin, 2011, Golier et al., 2003). Hence, all types of visual or verbal materials that may be perceptively or semantically linked to the traumatic event may be differently encoded and recollected than any other material, even emotional. Additionally, this phenomenon can be accompanied by increasing false recognitions of verbal or pictorial negative material. This bias for trauma-specific stimuli is also observed in attentional tasks and may interfere with another concomitant processing, such as in the Stroop task (Yiend, 2010). Two mechanisms are traditionally evoked, although not contradictory: first a facilitated detection based on an over-reactivity to traumatic cues (vigilance–avoidance model) and, second, a difficulty disengaging attention from threatening stimuli (attention–maintenance model). However, two recent meta-analyses report inconsistent results (Cisler et al., 2011, Kimble et al., 2009) suggesting that only a part of the PTSD population demonstrates this bias. A third hypothesis may conciliate this apparent inconsistence focusing on the bottom-up and top-down attentional systems (Bardeen and Orcutt, 2011). High levels of stress may impair some aspects of the attentional control particularly those which are top-down regulated, especially inhibition processes (Pineles et al. (2007) and shifting, part of executive functioning. Hence, a prominent theoretical perspective proposed the top–down inhibition of the amygdala by the ventromedial prefrontal cortex as a major mechanism that may be defective in PTSD. Bardeen and Orcutt (2011) reported greater difficulty to disengage from traumatic stimuli for individuals with prominent symptoms of posttraumatic stress. We can speculate that this deficit in attentional control or top-down process may affect memory.

Here, we examined this question by conducting two subsequent studies. The first study was conducted on a typical population of adolescents with the objective to assess top-down and bottom-up processes in emotion generation and their effect on attention and memory. This may enable us to validate our methodology and evaluate the “normal” influence of two different emotional processing: perceptual (or detailed-oriented) and conceptual (or based on the general meaning of the information). In the second study, we investigated these processes in patients with PTSD matched with controls, by distinguishing two subgroups, one with an attentional bias and another with no significant attentional interference, and their differential impact on memory. Two sets of negative emotional pictures were chosen from the internal affective picture system (IAPS; Lang et al., 2008) based on differences in subjective ratings relative to what induced the emotion, either perceptual details or a general interpretation of the scene. We studied both memory phases, i.e., encoding and explicit (recognition) retrieval. These two studies were approved by the local research ethics committee.