The power of imagination — How anticipatory mental imagery alters perceptual processing of fearful facial expressions

Abstract

Expectancies strongly shape our perception of the world and preconceptions about stimulus characteristics can even bias the sensory system for illusory percepts. Here we assessed with functional magnetic resonance imaging how anticipatory mental imagery of a mildly fearful face created a predictive bias that proactively altered perception of highly fearful faces and generated the “illusion” of reduced fearfulness. We found that anticipatory activation of the fusiform gyrus (FG) was modulated by the fearfulness of the imagined face. Further during anticipatory imagery, regulatory influences from the lateral and ventromedial prefrontal cortex on the FG primed the perceptual system for a subsequent misperception. This was achieved by increasing perceptual activation in higher-order brain regions for the evaluation of affective valence and contextual framing, while at the same time restricting bottom-up arousal and attention to fearful expressions. Anticipatory mental imagery may thus represent an effective antecedent strategy through which emotional perception can be significantly altered.

Introduction

Human survival critically depends on the ability to anticipate and prepare for likely future events in the absence of sensory stimulation (Bar, 2007, Holland, 1990, Ingvar, 1985). Central to this capacity is the ability to recur on mental representations of prior experiences or general preconceptions in higher-order brain regions, which limits the number of possible percepts and helps lower-level cortices to decide between competing percepts (Bar, 2007, Bar, 2009, Bouton, 1994, Engel et al., 2001, Frith and Dolan, 1997, Friston, 2005, Kveraga et al., 2007, Murray et al., 2004, Moulton and Kosslyn, 2009; see also Bar et al., 2006, Summerfield et al., 2006a, Summerfield and Koechlin, 2008).

However, if a subjective experience is indeed a composite of predictive information and sensory input (Petrovic et al., 2005), then the partial reliance on internal processing also bears that risk that mental representations can gain precedence over sensory input. In particular when preconceptions are strong, ‘predictive codes’ (Friston, 2005, Mumford, 1992) in higher-order brain regions may incorrectly ‘explain’ lower-level sensory information resulting in the generation of false or illusory percepts (Summerfield et al., 2006b). Evidence from brain imaging studies suggests that false expectations have the power to alter perceptual judgment, affective responses and neural processing in various stimulus modalities, including gustatory processing (e.g., Nitschke et al., 2006a), olfaction (e.g., De Araujo et al., 2005), vision (e.g., Petrovic et al., 2005), kinesthesia (e.g., Naito et al., 2007), and pain (e.g., Wager et al., 2004). One key finding of these studies was that misleading advance information modulated the perceptual experience by enhancing prefrontal top-down influences on category-specific sensory brain activation. For example, during face perception activation in the fusiform gyrus was found to be controlled by content-sensitive backward connections from various prefrontal regions, which determined whether illusory faces were detected in pure noise images (Li et al., 2009). Similarly, after subjects learned to associate a ‘placebo cue’ with a factual analgesic or otherwise comforting effect, increased prefrontal activation scaled with the attenuation of sensory-affective processing during perception of the aversive event and correlated with the concurrent decrease of negative affect (Keltner et al., 2006, Kong et al., 2006, Petrovic et al., 2005, Sarinopoulos et al., 2006, Wager et al., 2004, Watson et al., 2009; see also Benedetti et al. (2005) for a review of findings).

Although it is well established that false expectancies modulate the subjective experience of sensory-affective stimuli, the actual role of anticipatory neural processing in the generation of illusory percepts and in the reduction of negative affect has been widely neglected. One possibility is that crude forms of mental simulation operating prior to stimulus onset may contribute to the conscious expectancy of a certain stimulus and may act as a prime for the upcoming event and its relevant features (Bar, 2009, Ingvar, 1985, Holland, 1990, Moulton and Kosslyn, 2009, Rescorla, 1988, Rilling and Neiworth, 1987). In fact, a number of neuroimaging studies have demonstrated that anticipation activated partly the same or directly adjacent brain regions as those operating during actual stimulus perception (Bermpohl et al., 2006, Boly et al., 2007, Carlsson et al., 2000, Ploghaus et al., 1999, Porro et al., 2002, Nitschke et al., 2006b, Onoda et al., 2008). Moreover, there is evidence that the individual susceptibility to noxious somatosensory stimulation is related to fluctuations in both prestimulus activity and the anticipatory connectivity of networks relevant for somatosensory perception and its modulation (Boly et al., 2007, Ploner et al., 2010). Finally, it has already been demonstrated that the expectancy of a less noxious event than actually presented (i.e., a placebo stimulus) increased anticipatory activation in the prefrontal cortex (PFC), which correlated with both the effectiveness of the placebo treatment and the decrease of negative sensory-affective responses during stimulus perception (Sarinopoulos et al., 2006, Wager et al., 2004, Watson et al., 2009). Together, these previous findings converge to imply that (1.) the perceptual system may be part of a simulation network that can already shape sensory-affective representations during anticipation, and that (2.) prefrontal top-down influences during the anticipatory period may modulate sensory-affective responses in lower levels of the perceptual system. This could promote the generation of misperceptions and could also help to regulate fear-related arousal and negative affect during perception.

Here we investigated with functional magnetic resonance imaging (fMRI) whether voluntary anticipatory mental imagery of mildly fearful expressions has the power to proactively modulate perception of highly fearful faces and to generate the “illusion” of reduced fearfulness. Mental imagery or ‘seeing with the mind's eye’ holds the power to build up vivid internal ‘as if’ representations, which enable a mental simulation of actual visceral and emotional responses (Bensafi et al., 2007, Dadds et al., 1997, Kim et al., 2007, Kosslyn et al., 1996, Vrana and Lang, 1990). It has been proposed that mental imagery reflects the effects of prior knowledge on sensory processing areas in the absence of sensory input (Dolan et al., 1997, Frith and Dolan, 1997), and further allows a simulation of the future (Moulton and Kosslyn, 2009). Based on these considerations, we hypothesized that if anticipatory mental imagery is considered as a special case of ‘predictive coding’, then it should proactively modulate the subjective experience of fearful facial expressions. This would be in agreement with recent contemplations (Bar, 2007, Bar, 2009, Moulton and Kosslyn, 2009) and empirical evidence (Mechelli et al., 2004, Summerfield et al., 2006a) suggesting that voluntary mental imagery and ‘predictive coding’ of forthcoming perceptions share a common mechanism that enhances and directs prefrontal top-down influences on down-stream perceptual brain regions. We therefore made the prediction that anticipatory mental imagery would create a perceptual bias that is strong enough to proactively alter sensory-affective processing, which should predispose the perceptual system for a misperception of reduced fearfulness. Our study was designed to answer the following questions: (1.) How does the use of mental imagery modulate intensity-specific processing of fearful facial expressions during the expectancy phase? (2.) Which are the neurophysiological mechanisms that promote the impact of misleading advance information on perceptual judgment, if subjects mentally simulate the expected intensity of fearfulness during anticipation?

Participants performed an emotional evaluation task, in which they had to evaluate the fearfulness of human faces presented after an anticipatory period (see Experimental procedure). Trials were subdivided into three phases (anticipation, perception and rating phase) that were temporally jittered. Three cues predicted facial expressions of varying intensities of fearfulness (i.e., neutral, mildly and highly fearful; see Fig. 1 and Experimental procedure). Most cues correctly predicted the intensity of fearfulness of the upcoming expression. Only occasionally, a misleading cue predicted a less fearful expression than actually presented. This misleading condition was intended to induce the “illusory” perception of reduced fearfulness. Subjects were not informed about the misleading condition. We employed a between-subjects design with two matched groups, who were either explicitly instructed to prepare for the upcoming stimulus by imagining the expected facial expression during the complete anticipation phase (Imagery group) or who, as a control group, were simply informed that a cue predicted the intensity of the upcoming expression (NoImagery group). This allowed us to directly assess the mechanisms underlying anticipatory imagery of fearful expressions and its impact on subsequent perceptual processing.