Patients with bipolar disorder show a selective deficit in the episodic simulation of future events

Abstract

A substantial body of evidence suggests that autobiographical recollection and simulation of future happenings activate a shared neural network. Many of the neural regions implicated in this network are affected in patients with bipolar disorder (BD), showing altered metabolic functioning and/or structural volume abnormalities. Studies of autobiographical recall in BD reveal overgeneralization, where autobiographical memory comprises primarily factual or repeated information as opposed to details specific in time and in place and definitive of re-experiencing. To date, no study has examined whether these deficits extend to future event simulation. We examined the ability of patients with BD and controls to imagine positive, negative and neutral future events using a modified version of the Autobiographical Interview (Levine, Svoboda, Hay, Winocur, & Moscovitch, 2002) that allowed for separation of episodic and non-episodic details. Patients were selectively impaired in imagining future positive, negative, and neutral episodic details; simulation of non-episodic details was equivalent across groups.

Introduction

Recent work has examined autobiographical recollection in patients with bipolar disorder (BD). As in patients with major depressive disorder (MDD), these studies reveal that autobiographical memory in patients with BD is characterized by overgeneralized recall comprising primarily factual or repeated information as opposed to details specific in time and in place and definitive of episodic re-experiencing (Mansell and Lam, 2004, Mowlds et al., 2010, Scott et al., 2000, Tzemou and Birchwood, 2007). More recently, workers in the field of cognitive neuroscience have begun to examine the close relation between autobiographical recollection of past events and the simulation of future events (e.g., Spreng & Levine, 2006). In addition to identifying common cognitive processes including imagination (Addis, Pan, Vu, Laiser, & Schacter, 2009) and perspective taking (D’Argembeau & Van der Linden, 2004) that are involved in recollection and in future simulation, these studies have revealed a core neural network subserving both processes (Addis et al., 2009, Spreng et al., 2009). Critically, patients with BD show structural (for a review see Konarski et al., 2008) and/or functional (for a review see Drevets, Price, & Furey, 2008) brain changes in many of the same neural regions, including the hippocampus and prefrontal cortex, implicated in this network. At present, however, it is unclear whether deficits in the autobiographical recollection of past events among patients with BD extend to the related process of simulation of future events. The primary aim of the present study was to examine the ability of patients with this disorder to engage in prospective simulation of positive, negative and neutral events. A secondary goal was to examine the relation of simulation performance to several key clinical variables, including symptom severity and illness burden, shown previously to moderate autobiographical memory (AM) performance in patients with mood disorders (for a review see King et al., 2010).

Studies examining the relation between autobiographical recollection of past events and future simulation draw heavily on Tulving, 1985, Tulving, 2002) conception of autonoetic consciousness or mental time travel as the ability to travel either backwards or forwards in time and re-experience or pre-experience an event. More recently, Schacter and Addis (2007) proposed the constructive episodic simulation hypothesis that posits that while imagining future events individuals draw on memory of past events, recombining and elaborating details of past experience to simulate possible happenings. Hence, memory is an adaptive process (Bartlett, 1932, Suddendorf and Corballis, 2007), including the ability to predict future events with a fair amount of accuracy. Critically, simulation relies most heavily upon the ability to re-experience temporally, spatially and contextually specific episodic memory details that include happenings, sensations, perceptions, thoughts and feelings (Tulving, 2001). In contrast, semantic memory involves information that is context and time independent and is thought less crucial to future simulation. Of note, episodic and semantic memory have been shown to be dissociable neuropsychologically (Eslinger, 1998, Gardiner and Java, 1991, McKinnon et al., 2006) and in development (Fivush, 2010, Mäntylä, 1993, Parkin and Walter, 1992), as well as through imaging techniques (Levine et al., 2004, Svoboda et al., 2006).

In line with the constructive episodic simulation hypothesis, evidence gathered to date suggests that the neural substrates for recollecting the past and imagining the future show substantial overlap (Addis et al., 2007, Conway et al., 2003, Hassabis et al., 2007). For example, recalling the past or imagining the future engages both bilateral frontopolar cortex (Okuda et al., 2003) implicated in self-referential processing (Gusnard, Akbudak, Shulman, & Raichle, 2001) and medial temporal lobe (MTL) areas, including the hippocampus, involved in the retrieval of past events. In a recent study, increased activation in the left posterior hippocampus was associated with more vivid event descriptions for past and for future events (Addis & Schacter, 2008). Indeed, when compared to neurologically intact controls, patients with hippocampal damage recall fewer spatial details when asked to construct future events (Hassabis, Kumaran, Vann, & Maguire, 2007). A small number of neural regions, however, appear preferentially activated during recollection of past and of future events, including posterior visual cortex and parahippocampus (Addis et al., 2009, Addis et al., 2009). In one additional study, the left anterior hippocampus was more active during simulation of future imaged events than recollection of the past (Addis & Schacter, 2008).

Many of the same neural regions and cognitive processes implicated in future episodic thinking are impacted in patients with BD. Recollective memory impairments are among the most commonly reported cognitive deficits in patients with mood disorders (MacQueen, Galway, Hay, Young, & Joffe, 2002). Although results are inconsistent, volume loss has been reported in the hippocampus of patients with BD (Blumberg et al., 2003), and may contribute to this pattern of memory impairment (Yucel et al., 2007). Regions preferentially involved in self-referential processing and perspective taking, processes recently shown to be impaired in patients with BD (Cusi et al., 2010, McKinnon et al., 2010), also show volume loss and/or changes in functional activation, including the orbitofrontal cortex (Blumberg et al., 2006), subgenual prefrontal cortex (Haznedar et al., 2005), and the anterior cingulate (Farrow, Whitford, Williams, Gomes, & Harris, 2005). Of note, the dorsolateral prefrontal cortex also shows volume loss in BD (Coffman, Bornstein, Olson, Schwarzkopf, & Nasrallah, 1990). This area is strongly associated with working memory (Arts, Jabben, Krabbendam, & van Os, 2008). Finally, spatial processing (Shestyuk & Deldin, 2010) and imagery (Williams et al., 1996), key to both episodic recollection and future simulation, have been shown to be negatively affected in patients with mood disorders. Hence, it appears likely that the presence of structural and functional brain changes in patients with BD and their putative impact on cognitive function may alter related processes, including future episodic simulation. Indeed, many of these same neural regions and cognitive processes are thought central to autobiographical recollection (King et al., 2010), known to be impacted in BD and requisite to future simulation.

Few studies, however, have examined future thinking in patients with mood disorders. In one study, dysphoric individuals predicted that they would experience more negative events in their lives and generated these events more quickly than did controls (MacLeod & Cropley, 1995). In a similar experiment involving individuals diagnosed with MDD (MacLeod, Tata, Kentish, & Jacobsen, 1997), patients were less likely to predict future positive events than were controls; this finding was subsequently replicated in individuals with mixed psychiatric diagnoses who had attempted suicide (MacLeod et al., 1993, MacLeod et al., 2005).

In another study, Williams et al. (1996) examined the ability of patients with mixed psychiatric diagnoses (including MDD; none of the patients had a diagnosis of BD) who had attempted suicide to generate future event descriptions in response to positive, negative and neutral cue words. As predicted, patients generated less specific descriptions of both future and past events than did matched controls. Despite these early investigations, to date, no study has examined the qualitative characteristics of future episodic simulation in patients with BD, including number and type of events details generated. Here, we used a modified Autobiographical Interview procedure (see Addis, Sacchetti, Ally, Budson, & Schacter, 2009) to examine future episodic simulation in BD. In line with the constructive episodic simulation hypothesis, we hypothesized that this population would show poor episodic simulation of future events, given cognitive declines in constituent processes and changes in neural regions critical to past recollection and future simulation. Furthermore, given that AM deficits in BD appear to involve an inability to recall details specific in time and in place and definitive of episodic re-experiencing, we predicted that patients would experience greater difficulty simulating episodic than non-episodic details of future happenings.

In healthy populations, recall of emotional autobiographical events has been associated with heightened recollection compared to neutral events (D’Argembeau et al., 2003, D’Argembeau and Van der Linden, 2004, Schaefer and Philippot, 2005). In patients with BD, the impact of emotion on AM is less clear. Both Scott et al. (2000) and Tzemou and Birchwood (2007) showed that, relative to matched controls, patients with BD were equally overgeneral during recollection of past positive and negative events. By contrast, Mansell and Lam (2004) found that patients with BD were more overgeneral during recollection of negative compared to positive events. In order to examine whether emotion impacts future simulation in patients with BD, participants in the present study were asked to imagine future positive, negative and neutral events. Finally, in light of recent findings suggesting a relation between cognitive impairment, structural and functional brain changes, and clinical variables such as illness severity and past course of illness (e.g., Lebowitz et al., 2001, McKinnon et al., 2009, van Gorp et al., 1998), we examined the relation between these clinical markers and future simulation performance.