Very little is known about the underlying neurobiology that leads to persistent MdDS as there are no scientific studies that have established the underlying anatomical cause of this perceptual disorder.

Case series have shown that there are no inner ear or structural brain abnormalities associated with MdDS, which is not unexpected since the triggers that lead to persistent MdDS such as going on a cruise, flying in an airplane, or driving for long distances are unlikely to cause structural brain or inner ear damage [4]. However, MdDS may be a disorder of neuroplasticity in which entrainment to a periodically moving environment changes the functional state of the brain making it difficult to readapt to stable land conditions. Individuals with MdDS feel much better when back in motion again, such as getting back on the boat or driving in a car [4], [6]. The inability to shut off the mechanisms that become activated during motion exposure that prevent motion sickness (getting one's “sea-legs”) may be related to aging and hormonal factors, as MdDS is typically a disorder of middle-aged individuals and is more common in women [7].

The localization for this persistent motion illusion has not been previously sought with functional neuroimaging though previous theories have included “vestibular flashbacks,” and “internal modeling” of the external motion suggesting a central localization [6], [8]. The localization of a key hub of motion adaptation, one that can remain persistently active after the stimulus is removed, would represent a remarkable form of neuroplasticity that is relevant to brain adaptability in aging and that also plays a role in motion intolerance.

These data implicate the entorhinal cortex and amygdala in a human disorder of persistent motion perception and expand the role that EC/AG plays in human spatial memory. This is the first study to show metabolic abnormalities in patients with a chronic illusion of motion and provides information that may expand the relevance of limbic function to motion perception.

The entorhinal cortex is situated in the mesial temporal lobe anterior to the hippocampus. The medial entorhinal cortex receives highly processed spatial information from a broad swath of the neocortex, primarily from somatosensory association cortex and perirhinal cortex but with important connectivity to the medial prefrontal cortex and the amygdala. It is the main gateway of neocortical information entering the hippocampus [13]–[15]. ‘Grid cells’ located in the medial entorhinal cortex whose activity is highly correlated with spatial location, speed, and direction of heading are specialized cells that exhibit high directional selectivity and are primed to encode information about the current context of movement [16]. Grid cell maps are not rigid but are quickly modified by environmental context [17], [18].

Experiments done in humans playing virtual reality video games while undergoing fMRI scanning have revealed entorhinal cortex activity modulated by perceived speed and direction of movement [19]. Activity in the left hippocampal formation specifically activates when egocentric information is preferred for spatial navigation, whereas the right hippocampal formation favors an allocentric strategy [20]. Thus, the finding of the left entorhinal cortex activity in MdDS, a disorder of chronic self-motion perception, may relate to a network that is specialized for processing spatial information using an egocentric spatial reference and which normally functions to encode direction and speed specific information.

A distinguishing feature of MdDS is the reduction in the rocking perception with re-exposure to passive motion, a unique and unexplained feature among balance disorders [4]. In essence, the external motion appears to null the motion perception. A theory for this has only been proposed as an adaptation to the prior motion stimulus. A more specific theory proposed here relates to a key function of the entorhinal cortex as a major hub of generating network oscillations, specifically theta oscillations that drive the creation of place fields within the hippocampus, the key to spatial navigation [21]–[23]. Direct stimulation of the human entorhinal cortex during epilepsy surgery improves spatial navigation skills with positive learning trials being associated with a resetting of the phase of the theta rhythm in the hippocampus [24]. One theory for the nulling effect of re-exposure to passive motion in MdDS (such as returning to sea or driving a car) is that the frequency and amplitude of the incoming vestibular and somatosensory signals override or phase cancel out an underlying oscillating rhythm.

Of relevance to MdDS is that the entorhinal cortex is one of two brain areas that can generate sustained activity in the absence of continuous input; the other being the prefrontal cortex to which it is highly connected [25]–[27]. Single pulse stimulation of entorhinal cortex can generate a recurrent loop of activity within the entorhinal cortex itself, which then spreads into the hippocampus when past threshold [28]. Periodic stimulation can also drive baseline entorhinal cortex faster or slower, depending on the period of the stimulation. One elegant study on in vitro brain slices showed that depolarizing electrical stimulation to the entorhinal cortex given in four second pulses raises its baseline firing rate, whereas hyperpolarizing pulses given in six second pulses gradually reduces the baseline firing rate to the point of stopping the activity [29]. Thus, baseline entorhinal cortex activity can potentially be modified by a change in the periodicity of its neuronal inputs making it is a plausible substrate for entrainment by external stimuli.

The amygdala activity seen in these data is consistent with the dependence of the entorhinal cortex on amygdala activity for efficient information transfer into the hippocampus [30]. Transfer of sensory information into the entorhinal cortex largely comes through the perirhinal cortex but studies in rat brain slices have shown that individual stimulation of perirhinal cortex is inefficient in spreading activity into the entorhinal cortex. However, concurrent stimulation of the amygdala promotes spread of activity from perirhinal cortex to the entorhinal cortex and subsequently into the hippocampus, a phenomenon that may be the basis for the role of motivational and emotional variables in memory consolidation [31]

The role of the prefrontal cortex in the generation of MdDS may be due to its strong functional and anatomical connections to both the entorhinal cortex and amygdala, specifically being shown in animal models to be inhibited by the entorhinal cortex while exerting regulatory control over both these areas [26], [32]. The reduced prefrontal connectivity seen in these data, particularly in the setting of increased connectivity with motion processing areas in posterior parietal, temporal, and occipital areas may together contribute to the increased activity with the entorhinal cortex and amygdala and contribute to a poorly regulated network where there is enhanced transmission of motion information into limbic areas with less regulatory control. Only connectivity between the EC/AG and other brain regions were investigated in this study. It is possible, however that connectivity between these areas (e.g. prefrontal and posterior parietal cortex) is also important in the generation of MdDS symptoms.

One phenomenon in MdDS that suggests a very strong role for structures involved in memory consolidation is that many patients experience a spontaneous re-emergence of their rocking dizziness even years after their original MdDS subsides [4]. The role of the amygdala may be relevant to why some patients who have an initial episode of MdDS that resolves tend to have spontaneous recurrences of the rocking sensation during periods of stress [7]. In addition, when patients experience recurrent episodes of MdDS, the episodes tend to become longer [4]. This suggests that functional connections that are created during the original period of habituation to background sea or air conditions can be strengthened with time and may develop a lower threshold for activation with subsequent stimulation.

Because of the inherent confound of concurrent mood and anxiety disorders which complicate chronic illness, we found it necessary to take extra steps to be sure that differentially active brain areas in MdDS were not simply due to anxiety or depression. The HADS is a well-validated screening tool of seven items of anxiety and seven items of depression [33]. We used scores on the HADS anxiety and depression subcomponents as nuisance covariates to remove signal that varied with high HADS subscores. However, in order to determine whether the peak voxels in the contrast images may have occurred even near regions implicated in anxiety and depression and may have been considered distinct simply because of thresholding, we used a multivariate regression analysis to identify brain areas that were specifically associated with high anxiety and depression scores. Using this analysis, classical brain areas that have been previously reported to be under/overactive in anxiety and depression were revealed but did not overlap with any areas seen in the primary analysis. These included the subgenual anterior cingulate cortex activity that correlated with depression [34]–[36] and dorsal midbrain and anterior temporal lobe activity correlating with anxiety [37]–[39]. There are no resting state FDG PET studies in anxiety disorders that have shown abnormalities in baseline metabolic activity in the entorhinal cortex or amygdala but amygdala activity is inducible in many anxiety subtypes when measured with fMRI [40], [41]. Baseline glucose metabolism has been found to be higher in the left amygdala in depressed patients when analysis is limited to a region of interest over the amygdala [42]. Although it is possible that the amygdala activity in our data is related to baseline depression in MdDS subjects, the entorhinal cortex activity is not explained by previous imaging findings in mood and anxiety disorders.

Because it was still possible that the findings of enhanced entorhinal cortex and amygdala activity were related to higher baseline mood symptoms that were not fully measured with the HADS questionnaire, we also undertook a functional connectivity analysis to determine whether there were different patterns in connectivity that could relate to motion processing pathways. These connectivity differences appeared to be specific to the posterior parietal and occipital areas that process visuospatial information, with connectivity being enhanced in subjects with MdDS despite overall connectivity patterns being reduced between frontal and temporal areas.

The data presented here expand the relevance of the metabolic activity and connectivity of limbic structures to a human disorder of continuous perception of self-motion. Despite MdDS being considered a rare disorder, the study of a condition where the initial adaptation to background oscillating conditions do not successfully readapt back to stable conditions would still reflect a pathway relevant to motion adaptation. These pathways would be easier to measure in MdDS patients because the symptoms are stable and persistent, unlike the transient feelings of rocking felt by most individuals after travel that can dissipate within minutes. Although the controls were matched as well as possible given age, sex, and handedness, a more ideal control group would have been one that was also matched for the motion exposure. From a practical standpoint, this was not possible since there are infinite combinations of sea, air, and land conditions that may have been relevant to why the MdDS subjects had developed their symptoms.

Figure S1

Figure S2

Table S1