Although anorexia nervosa (AN) is a critical and potentially mortal illness (Mustelin et al., 2016), its etiology is still unclear (Kaye et al., 2013). Recent neurobiological studies consider AN as the outcome of dysfunctions in the neuronal processes related to appetite and emotionality (Kaye et al., 2009, 2013).

In their reviews, Kaye et al. (2009, 2013) suggest that AN patients are characterized by a dysregulation in the anterior ventral striatal pathway that may create a vulnerability for dysregulated appetitive behaviors. The high level of self-control in individuals with AN—produced by an exaggerated dorsal cognitive circuit functioning—allows them to inhibit appetite.

This model, even if very influential and able to provide clear suggestions for therapy, still is not able to answer a critical question: What is behind body image distortion in AN? As noted by Kaye et al. (2013): “This may be the most puzzling of all AN symptoms, in part because AN individuals feel fat but tend to have normal perceptions of other people's bodies” (p. 117).

A possible path for providing an answer to this question is to investigate the role of altered monoamine neural modulation in AN (Haleem, 2012).

The term “monoamine neurotransmitters” refers to the particular dopamine (DA), noradrenaline (NAD), adrenaline (AD), and serotonin (5-HT) neurotransmitters that are released from neurons in both the brain and peripheral nervous system (Kaye et al., 1984) and that affect a variety of psychobiological factors including hunger, anxiety, impulsivity, perception, and memory. These neurotransmitters have been extensively investigated, and different studies suggest a significant decrease in AN patients when compared to normal subjects (Kaye et al., 2013). In particular, the 5-HT and DA systems have a significant impairment in AN patients (Kaye et al., 2009), with possible effects on satiety, impulse control, and mood (5-HT), and aberrant rewarding effects (DA) of motivation and food (Kaye et al., 2013).

Although the possible role of dopamine in AN is still controversial (O'Hara et al., 2015; Peng et al., 2016; Södersten et al., 2016), a number of studies evidenced an altered serotoninergic activity in AN (Kumar et al., 2010; Calati et al., 2011; Jean et al., 2012; Chen et al., 2015) and demonstrated the role of the 5-HT receptors located in the hypothalamus in food intake and body weight control (Compan et al., 2012; Haleem, 2012). In this view, the 5-HT impairment may have a clinical role in explaining the insufficient food consumption in AN. As underlined by Compan et al. (2012): “The brain 5-HT system is central in the control of food intake and particularly in eating disorders…Accordingly, environmental changes (stress) could alter the adaptive decision-making concerning feeding. If the adaptive response to stress depends on the 5-HT system, eating disorders could thus emerge when 5-HT neurons reach the limit of their adaptive capacities.” (p. 723). Additionally, early-life stress, a risk factor for eating disorders (Su et al., 2016), induces persistent changes in 5-HT receptors and transporter (Bravo et al., 2014).

However, recent studies also suggest a link between 5-HT and memory dysfunctions, creating a possible bridge between serotonin disturbances, impaired body memory, and body image disturbances.

Our experience of the body is not direct (Figure 1), but it is mediated by perceptual information, recalibrated through stored information (body representations) and influenced by internal information—proprioception, interoception, and vestibular input (Blanke et al., 2015; Pazzaglia and Zantedeschi, 2016).

In this view, body image distortion can be seen as a multidimensional construct (Figure 1) that, according to recent neuroimaging studies, includes three different components: affective, cognitive, and perceptive, (Gaudio and Quattrocchi, 2012). The affective and cognitive components of body image distortion are widely accepted and related both to socio-cultural issues (van den Berg et al., 2002; Tylka, 2011; Swami, 2015)—internalization of an ideal body figure (Fitzsimmons-Craft et al., 2016), social body comparison (Andrew et al., 2016), and self-objectification (Dakanalis et al., 2016a)—and/or to brain dysfunctions (Sato et al., 2013; Suchan et al., 2015; Esposito et al., 2016)—alterations of the prefrontal cortex, the insula and the amygdala (Dakanalis et al., 2016b).

The interest for the perceptual component of body image distortion is more recent and related to the outcomes of different functional magnetic resonance imaging (fMRI; Mohr et al., 2010; Suchan et al., 2013). Gaudio and colleagues summarizing these studies in a recent systematic review (Gaudio et al., 2016) conclude “that several brain regions could be involved in body image disturbances and may sustain an impaired integration between real and perceived internal/external state of one's own body in AN patients” (p. 582). For example, neuroimaging studies have demonstrated both functional and structural more marked alterations in visual areas, with anomalies in body-image processing related to self but not to others (Uher et al., 2005; Sachdev et al., 2008). A specific association between therapy-related changes and modulation of the BOLD signal in these areas suggests that this type of distortion of the body's configuration is based on visual contributions (Vocks et al., 2011). Specifically, a recent study showed that AN patients' perception of their own body is more easily malleable by exposure to round figures as compared to controls (Cazzato et al., 2016). Nevertheless, another systematic review underlined a multisensory impairment of body perception in AN that goes beyond visual misperception and involves tactile and proprioceptive sensory components and suggests a critical role of memory in these processes (Gaudio et al., 2014). As suggested by two recent reviews (Longo, 2015; Martijn et al., 2015) and underlined by Fuchs (2012): “The body is not only a structure of limbs and organs, of sensations and movements. It is a historically formed body whose experiences have left their traces in its invisible dispositions…Sensations or situations experienced by the lived body may function as implicit memory cores which, under suitable circumstances, can release their enclosed memories.” (p. 20).

An emerging etiological model of AN, that focuses on body memories is the “Allocentric Lock (AL) Hypothesis” (Riva, 2012, 2014; Riva et al., 2015).

This theory suggests, that AN is the outcome of a primary disturbance in the way the body is “experienced” and “remembered.” Specifically, individuals with (or developing) this disorder may be locked to an allocentric (third person) disturbed memory of their body that, independently of its causes, is not more updated by experiential data, even after a successful diet and/or a significant weight reduction (Figure 1—for a broader review see: Riva, 2007, 2011, 2012, 2014; Gaudio and Riva, 2013; Riva et al., 2015; Dakanalis et al., 2016b).

The article highlights some of the important studies discussing the effects of 5-HT markers (i.e., receptors and transporter) on memory dysfunctions, which may play a role in clarifying the link between decreased serotonin transmission, body memory, and body image disturbances.

A wealth of experimental animal studies demonstrate the role of the serotonin systems in memory and its dysfunctions, even if their role is still poorly understood (Meneses et al., 2011; Meneses, 2013; Gasbarri and Pompili, 2014; Gasbarri et al., 2016; Meneses and Gasbarri, 2016). It is beyond the scope of this paper to describe the results of all these studies (for an in-depth analysis please refer to Meneses et al., 2011; Roberts and Hedlund, 2012; Meneses, 2013; Glikmann-Johnston et al., 2015; Meneses and Gasbarri, 2016). In this context, we will focus only on the studies more relevant to our discussion.

The studies discussed above, even if largely based on animal research, suggest a link between the 5-HT systems and the memory processes—encoding/storage, consolidation, and retrieval/reconsolidation (see Figure 2)—involved in the allocentric lock. Furthermore, 5-HT effects appear to be modulated by stress (Baratta et al., 2016) and estrogen (Epperson et al., 2012), and have a stronger influence on working and short-term memory than on long-term memory (Hritcua et al., 2007).

In relation to encoding and storage, the study by Lemogne et al. (2009) suggested that the 5-HTTLPR polymorphism influences the visual perspective used for storing positive memories during life stress. Specifically, individuals with at least one low or long _G allele used more a third person perspective in episodic memory, facilitating the storing of an allocentric memory of their body. More, the results of the study by Gonzalez-Burgos et al. (2012) showed that serotonin depletion impaired the egocentric working memory performance while no effects were found for the allocentric one.

In relation to the consolidation process, it is well known that in humans, it is impaired by a decrease in serotonergic neurotransmission (Sivamaruthi et al., 2015). Animal studies also have found that a decrease in 5-HT reduced conditioned responses in both short and long-term memory (Gonzalez et al., 2013), while its availability improved the stabilization of emotional memories (Baratta et al., 2016), if preceded by a history of repeated stress exposure (Ohmura et al., 2015).

The reviewed studies also suggested a critical role of 5-HT in memory retrieval and reconsolidation. In humans, serotonin receptor 5-HTR_2A gene polymorphism was associated with a significant impairment in memory retrieval (de Quervain et al., 2003; Sigmund et al., 2008). Also, if we focus on the computational network model used to investigate 5-HT modulation on spatial working memory (Cano-Colino et al., 2014), its results emphasized a critical role of 5-HT in the inhibition of currently irrelevant or unwanted information (Marsh et al., 2012). According to the authors: “Increasing levels of tonic [5-HT] …favored spatial working memory by suppressing unwanted false memories” (p. 2459). Sarkisyan and Hedlund, discussing the results of their study (Sarkisyan and Hedlund, 2009), suggested: “The 5-HT₇ receptor might be important in the formation of associations with the network of memories and the compilation and correlation of these memories with changes in the environment” (p. 29).

Recalling information from long-term memory usually requires the selection of a specific target from a wider set of targets competing for access. Taken together, these effects suggest a role of serotonin in inhibiting unwanted targets (proactive interference) that can be impaired during decreased serotonin transmission. Two studies with monkeys confirmed this interpretation (Clarke et al., 2004, 2007): selective serotonin depletion of the orbitofrontal cortex impairs the ability to switch between visual stimuli when responding on a serial discrimination reversal task. Specifically, as the authors noted: “The failure of 5-HT-lesioned monkeys to cease responding to the previously correct stimulus was due to an inability to disengage from that stimulus” (p. 24). Furthermore, a recent study also points out the role of serotonin in memory reconsolidation (Nikitin et al., 2016): in snails, the 5-HT receptor antagonist induced the disruption of memory reconsolidation related to previously conditioned food aversion.

Finally, if we focus on the brain systems involved in the allocentric computations (Ekstrom et al., 2014)—hippocampus, retrosplenial cortex, and parahippocampal cortex—we can find a significant serotonergic innervation affecting the proliferation and activity of their cells.

For example, in mice, 5-HT receptors induce spine growth in the CA1 (Restivo et al., 2008), an area of the hippocampus involved in spatial autobiographical memory and in the development of allocentric view independent representations. It is also well-known that serotonin synthesis has a positive regulatory factor on the granule cell layers of the retrosplenial cortex, a brain area involved in transforming allocentric representations into egocentric ones (Vann and Aggleton, 2005; Vann et al., 2009) by improving their proliferation (Richter-Levin and Segal, 1990; Brezun and Daszuta, 2000). A decreased serotonin transmission may impair these areas, disrupting their functions. A study of 5-HT7 receptor-deficient mice (5-HT7 –/–) supported this hypothesis (Sarkisyan and Hedlund, 2009): in a spatial memory task the mice demonstrated an impaired allocentric spatial memory whereas egocentric spatial memory remained intact. Also, different recent neuroimaging studies with AN patients revealed significant brain dysfunctions in the key areas involved in the allocentric computation (Riva and Gaudio, 2012; Gaudio and Riva, 2013).

In conclusion, this review suggests that on one side, 5-HT systems modulate memory and its dysfunctions; on the other side, a decreased serotonin level impairs the different memory processes—encoding/storage, consolidation, and retrieval/reconsolidation—involved in episodic and autobiographical memory (see Figure 2). Specifically, serotonin disturbances:

- Facilitate the encoding of allocentric (from outside) episodic memories;

If we discuss these data within the interpretative frame suggested by the Allocentric Lock Hypothesis, we can hypothesize that AN patients:

Are more able to store and consolidate negative autobiographical memories, including those of their body, in allocentric perspective. As demonstrated by Eich et al. (2009, 2012), these memories produce a significant reduction in one's cortical representations of the physical self. As clarified by the authors (Eich et al., 2012): “When we choose to relive past events from a perspective outside our body, we shut down the neural circuitry in the insula that is central for monitoring our bodies' internal states” (p. 177). A first support for this hypothesis is given by the multisensory impairment of body perception existing in AN patients, suggesting that their bodily experience is shaped by sensorimotor/proprioceptive memory (Gaudio et al., 2014).

In conclusion, the data accumulated in this review suggest a possible link between serotonin dysfunctions and body image disturbances in AN: the impossibility of updating a disturbed body memory using real time experiential data—I'm locked to a wrong body stored in long term memory—pushes AN patients to control body weight and shape even when underweight. This view is in agreement with a recent hypothesis that describe anorexia as a disturbance of the self (Amianto et al., 2016) specifically associated “with spatial functioning possibly related to experiencing one's own body as an integrated aspect of the self, and temporal functioning possibly related to integrating the self in a coherent narrative over time.” (p. 7). More, it is also in agreement with the free energy framework suggesting that pre-existing mental models shape current perception (Friston and Kiebel, 2009). According to this vision the brain maintains hypotheses of the causes of sensory input predicting inputs, which are then compared with actual sensory input. In this view body image disturbances may be the outcome of dysfunctional predictive mechanisms (Seth et al., 2012; Di Lernia et al., 2016).

However, it is also true that this review is largely based on animal studies. Due to this, there exists a theoretical gap between the outcome of these studies and the possible effects of serotonin in humans. Further, studies are needed to explore and clarify the link between human serotonin dysfunctions, body image disturbances, and the causes of AN.

GR conceived, collected, analyzed, interpreted the data, and gave his final approval of the version to be published.

This paper was supported by the PRIN 2015 project “Unlocking the memory of the body: Virtual Reality in Anorexia Nervosa” (201597WTTM).