Experiments were carried out in 40 subjects (age: 38.38 ± 10.05 years; females (n = 17): 36.00 ± 9.82 years; males (n = 23): 40.13 ± 10.06 years), who signed a written informed consent. Considering the side with higher EMG activity as the hypertonic side and the side with lower EGM activity as the hypotonic side, subjects showed a baseline EMG asymmetry greater than 20%. The asymmetry was evaluated as follow:

All subjects were native Italian speakers and unaffected by neurological or psychiatric disorders. The study was conducted in accordance with the Declaration of Helsinki and the experimental protocol was approved by the Ethical Committee of the Pisa University (endorsement 39/2022).

During clenching, right and left masseter EMG activities were recorded by a K6-I MyoTronics system (sample rate: 720 Hz, cut-off frequency: 15 Hz, notch filter: 50 Hz) using Duotrode Ag/AgCl electrodes (inter-electrode distance: 19.5 mm, MyoTronics, Seattle, WA, USA), placed as described by De Cicco et al. (2014). Automatically filtered EMG traces were fully rectified, and the values of burst duration and average EMG activity were computed. The clenching effort (2–4 s) began and ended upon experimenter’s request.

Subjects received 15 min of transcutaneous electrical nerve stimulation (TENS) of trigeminal motor branches (Nnoaham and Kumbang, 2008), using four couples of electrodes, positioned on both sides at the level of the incisura sigmoidea and the submental region. Biphasic (cathodal/anodal) current pulses (0.54 ms duration, 21–25 mA intensity), delivered by two I.A.C.E.R. stimulators (Martellago, Venice, Italy), induced repeated contractions of the masseters and mandible depressor muscles. The stimuli for the left and the right sides were adjusted to achieve symmetric muscle activation, with the frequency set at 40and 0.618 Hz for mandible depressor and elevator muscles, respectively. These patterns led to an alternating contraction of the masseters and a tonic contraction of depressor muscles, resulting in small amplitude mandibular movements (1 mm). After TENS, the mandibular resting posture was lower. A dental impression was then obtained by placing a self-hardening material between the dental arches. When the subject wore this dental impression during a clenching effort, the EMG asymmetry was greatly reduced or abolished.

After creating the dental impressions (imprint), all subjects participated in two consecutive reading sessions to assess their oculomotor performance. In the first session, the subject wore the dental impression (corrected condition), while in the second session, their dental arches were in direct contact (habitual condition). It has been observed that simply repeating the reading task enhance reading ease, leading to shorter total reading time and reductions in both the frequency and the duration of fixation periods (Hyönä and Niemi, 1990; Raney and Rayner, 1995; Raney et al., 2000). Therefore, we chose to begin the experiment under the occlusal condition that, based on our hypothesis, was supposed to result in the best reading performance. In this way, the lack of improvement or a worsening of the reading performance in the second test could have been attributed to the effects of dental impression removal.

No outlier rejection was applied to the population analyzed. Oculomotor data were preliminary tested for distribution normality by the Kolmogorov–Smirnov test and by the Shapiro–Wilk test. The results of this screening have been shown in Table 1. Since the more demanding test (Shapiro–Wilk) indicated a significant deviation from normality (Table 1A), a Log10 transform was applied to the data. The transformation gave rise to a normal distribution for all the oculomotor parameters analyzed (Table 1B). Data were represented as mean ± standard deviation (SD). Between condition comparisons were performed by paired t-test. Correlations between variables were investigated by linear regression analysis. In addition to these classic parametric statistics, more robust tests less biased by data distribution were performed for both regression analysis (rank Spearman, Gamma and Kendall Tau correlation) and between-condition comparisons (Wilcoxon test). Most of statistical analyses were performed with Statistical Package for Social Sciences (SPSS, version 20). For non-parametric correlations the software STATISTICA (version 6.0) was utilized. The level of statistical significance was set at p < 0.05.

In the habitual condition, the EMG activity recorded on the hypotonic and hypertonic side corresponded to 61.9 ± 40.0 μV and to 126.0 ± 43.0 μV, respectively (t-test: p < 0.0005). Following occlusal correction, these EMG values were rebalanced (hypotonic: 143.0 ± 55.4 μV, hypertonic: 147.2 ± 51.0 μV, t-test: p = 0.090). The increase in EMG activity was significant for both sides (t-test: p < 0.0005). These modifications led to a significant reduction of the absolute value of EMG asymmetry in the corrected condition (from 76.8 ± 42.4% to 12.0 ± 9.8%, t-test: p < 0.0005). All these findings were confirmed by the Wilcoxon test. Basal values of EMG asymmetry and their reduction after occlusal correction were not correlated with subjects’ age. Linear regression analysis indicated that the (habitual - corrected) changes in EMG asymmetry were significantly correlated with those observed in the EMG activity recorded on the hypotonic side (Figure 1A). This finding was confirmed by non-parametric models of correlation (see Methods). Differently, no correlation was observed with the EMG changes recorded on the hypertonic side (Figure 1B).

A summary of the different oculomotor metrics recorded under corrected and habitual condition is displayed in Table 2. The corrected condition was associated with a lower fixation number compared to the habitual one. The same held true also for the number of fixations occurring per time unit. On the contrary, fixation duration was higher in corrected than in the habitual condition. Changing from the corrected to the habitual condition, no significant difference was observed in the total reading time. However, as shown in Figure 2, individual subjects could either show increases (circles) or decreases (dots) in this parameter. We will refer to these groups of subjects as Habitual Slower (HS) and Habitual Faster (HF), respectively. It is of interest that the increase in number of fixations characterized both HF (corrected: 2.02 ± 0.16 Log₁₀(Nos.), habitual: 2.11 ± 0.16 Log₁₀(Nos.), t-test: p = 0.019, Figure 3A) and HS (corrected: 2.02 ± 0.17 Log₁₀(Nos.), habitual: 2.11 ± 0.13 Log₁₀(Nos.), t-test: p = 0.001, Figure 3B) subjects. At variance, the decrease in fixation duration when passing from corrected to habitual condition was found among HF (corrected: 2.59 ± 0.20 Log₁₀(ms); habitual: 2.43 ± 0.19 Log₁₀(ms), t-test: p < 0.0005, Figure 3C) but not among HS subjects [corrected: 2.53 ± 0.24 Log₁₀(ms); habitual: 2.50 ± 0.21 Log₁₀(ms), t-test: p = 0.310, Figure 3D]. Nearly identical results were observed when data were compared by the Wilcoxon test. No differences were found between HF and HS subjects in terms of age, EMG asymmetry, oculomotor parameters and reading times evaluated in both corrected and habitual conditions.

When subjects were tested in the corrected condition, linear regression analysis did not show any significant correlation between total reading time/oculomotor measurements and the corresponding level of EMG asymmetry. Differently, in the habitual condition, the number of fixations (Figure 4A) and their duration (Figure 4B) were positively and negatively correlated with the EMG asymmetry, respectively. A positive correlation with EMG asymmetry was also found when the number of fixations was divided by the total reading time (number of fixations per second, Figure 4C). Differently, no significant correlation was observed between the total reading time and the EMG asymmetry (R = 0.212, p = 0.189). All these correlations were confirmed by applying non-parametric models. As shown in Figure 5, in both the corrected (Figure 5A) and the habitual (Figure 5B) conditions, fixation number and fixation duration were strongly (and negatively) correlated. While the total reading time was independent upon the fixation frequency, it was positively correlated with the fixation duration in both the corrected (Figure 5C) and the habitual (Figure 5D) conditions. These finding were confirmed by non-parametric regression models.

The differences in the number of fixations (Figure 6A), in fixation duration (Figure 6B) and in the frequency of fixations per second (Figure 6C) observed by switching from the corrected to the habitual condition (Log habitual - Log corrected) were influenced by the relative changes in (habitual-corrected) EMG asymmetry, as indicated by linear regression analysis. This observation was confirmed by non-parametric correlations.

The main limitation of the present study is the lack of a reversal in the order of the two occlusal conditions (corrected, habitual) under which the reading task was performed. The lack of reversal poses the problem of how to reasonably dissect the effects of the condition from those of repeating twice the same test (the corrected condition was tested before the habitual condition). In this respect, the data present in the literature allow to draw some conclusion despite the methodological flaw. Indeed, there is a large body of evidence that the second reading of a text results in a total reading time reduction and in a decrease in both the frequency and the duration of fixation periods (Hyönä and Niemi, 1990; Raney and Rayner, 1995; Raney et al., 2000). Reading difficulties, such as dyslexia, are typically characterized by the opposite pattern, i.e., an increase in both the number and the duration of fixation and an extension of the total reading time (Hutzler and Wimmer, 2004). This pattern has been observed also in patients with macular degeneration (Ktistakis et al., 2023), Alzheimer’s disease (Lueck et al., 2000; Fernández et al., 2013, 2015) and aphasia-based alexia (Hux et al., 2024). So, in the present study, the lack of an improvement or a worsening of oculomotor parameters during the second session suggests a negative effect of the habitual condition on text reading. Indeed, according to our hypothesis, the balanced (corrected) occlusal condition tested first should have been that associated to a better oculomotor control. Another point that allows to distinguish the effect of occlusal condition from those of task repetition is the fact that, in the former case, the changes observed in oculomotor behavior should be correlated with the associated changes in EMG asymmetry. This was indeed observed in the present study, as shown in Figure 6.

Finally, another limitation of the study was the lack of measuring the saccade amplitudes, whose acquisition would have allowed a better understanding of the changes in oculomotor behavior elicited by changing the occlusal condition. It has to be pointed out, however, that given the fixed length of the read text and the relative small number of regressions performed during reading (Rayner, 2009; 10–15% of the time), the changes in saccade amplitude can be expected to be strongly (and negatively) correlated to those in saccade (fixation) frequency.

Findings from this study suggest that oculomotor behaviour is modified by the occlusal condition.

In our study, during the second reading session, we did not observe an average decrease in the total reading time. Conversely, we noticed an increase of the frequency of fixations and a decrease in the average duration of fixations. This suggests that removing the imprint for the second reading session might have partially negated the typically positive impact that a second reading has on oculomotor metrics, consisting in a decrease in fixation frequency and duration. This implies that the influence of occlusal condition on reading performance may be significant enough to override the decrease in fixation frequency expected from the familiarity and repetition of the reading task. This hypothesis is supported by the finding that the increase in fixation frequency, induced by switching from corrected to the habitual condition, was positively correlated with the corresponding increase in EMG asymmetry during clenching. Since the duration of fixations was on the average reduced, as would have been expected during a second repetition of the reading task, it could be proposed that occlusal disharmony is hampering the sequence of eye movements more than the processing of visual information, which determines the duration of the fixation period. In some subjects, fixation duration showed little changes or a frank increase switching from corrected to habitual condition: consequently, they were reading faster with the imprint interposed between the arches. In this restricted population, the occlusal disharmony has not only increased the number of fixations but has also counterbalanced or overwhelmed the decrease in fixation duration expected by task repetition.

Beyond the changes in oculomotor metrics elicited by changing the occlusal condition, it has to be stressed that within our population oculomotor metrics in habitual occlusion are related to the trigeminal asymmetry: subjects with larger asymmetries show in fact higher number of fixations of shorter duration with respect to less asymmetric subjects. In other words, trigeminal asymmetry seems to determine the reading pattern of the individual subjects. This gives rise, within the population, to a trade-off between fixation duration and frequency, with an inverse relationship between the two oculomotor metrics. Higher number of fixations is indicative of a reduced saccadic amplitude and could be attributable, in principle, to a reduction in the span of text that can be processed by the reader, leading to a shift in reading dynamics/patterns characterized by an increased number of fixations (i.e., a reduced saccade amplitude) with shorter durations. Higher numbers of fixations and reduction of their duration with respect to normal controls has been also observed in offspring of late-onset Alzheimer’s disease (O-LOAD), as reported by Fernández et al. (2022). The authors concluded that “This eye movement pattern could be considered an early marker of oculomotor impairment.” They claimed that the control group processed and integrated the full meaning of a sentence through larger saccades and spending more time on the fixated words, whereas the O-LOAD group had difficulties in extracting information from the text, being constrained to a reading pattern characterized by shorter saccades and less time on the fixated word. In other words, the O-LOAD subjects could not analyse the same span of words as the control group.

The reading difficulties elicited by occlusal disharmony could be linked to a diminished capacity in processing visual input coming from the parafoveal region, which is crucial for efficient reading (Schotter et al., 2012): this would explain the higher fixation number and reduced fixation duration observed in our study. A parafoveal vision decline was observed in older readers (Payne and Federmeier, 2017). Indeed, these subjects showed a reduced ability to quickly combine the meanings of words seen in their peripheral and foveal vision (Payne and Federmeier, 2017). This could be related to neuroanatomical and functional changes induced by aging, particularly in the frontal cortex (Fabiani, 2012). Interestingly, in the presence of a trigeminal sensory-motor imbalance, a reduced performance was observed in the Spinnler-Tognoni test, consisting in the retrieval of target numbers within numerical matrices that must be scanned line by line (Spinnler and Tognoni, 1987). When the asymmetry was corrected by bite/oral prosthesis placement, it was observed an improvement in scanning velocity of the matrix, as well as in the speed of target retrieval, without any increase in error rate. These results contradicted the well-known trade-off between speed and accuracy (Fitts, 1954) and could be well explained by an increase in the subject’s ability to extract parafoveal information. This results in a higher number of items that can be processed simultaneously, allowing a faster execution of matrices scanning and targets retrieval. Based on these findings, the decrease in fixation duration following imprint removal could be regarded somehow differently than an improvement, in reading performance due to the repetition of the task. It could represent a change in reading pattern imposed by a reduced capability of processing visual information following the insurgence of a condition of occlusal disharmony. Further investigation is needed to clarify this issue.

Some of the central connections of trigeminal system may potentially account for an occlusal interference with oculomotor behaviour (Zhang et al., 2011; Liang et al., 2017). Indeed, trigeminal input reaches RF (Langer and Kaneko, 1990) and vestibular nuclear (Buisseret-Delmas et al., 1999) regions, as well as the nucleus prepositus hypoglossi (McCrea and Baker, 1985; Buisseret-Delmas et al., 1999) and the superior colliculus (Ndiaye et al., 2000), which have part in oculomotor control. These anatomo-functional relationships could account for pathological interference of mandibular movements with eye movements (Pratt et al., 1984; Zhang et al., 2011), as well as for the apparent interference existing between occlusal condition and ocular vergence (Marchili et al., 2016; Caruso et al., 2018). In principle, an asymmetry in the trigeminal input may complicate oculomotor control by affecting the activity of the structures mentioned above, resulting in the need of a larger number of saccades to read the same amount of text. Although we cannot completely reject this hypothesis, it must be pointed out that the observed changes in oculomotor behavior resembles to an adjustment in reading dynamics rather than to an oculomotor impairment.

Subjects with asymmetric sensorimotor trigeminal activity during clenching were characterized by anisocoria. Giving the established relationship between pupil size and LC activity, this finding suggests that in our sample the trigeminal asymmetry may lead to an asymmetry in the basal LC activity. The pathways linking proprioceptive trigeminal activity to LC activity have been recently reviewed (De Cicco et al., 2017). Fibres from the mesencephalic trigeminal nucleus reach the LC directly or through the reticular formation. In turn noradrenergic fibres from the LC contact mesencephalic trigeminal ganglion cells (Takahashi et al., 2010). Few LC axons enter the trigeminal motor nucleus where they enhance motoneuronal excitability (Samuels and Szabadi, 2008). These feedback projections may potentially enhance proprioceptive trigeminal and LC activity during chewing, giving rise to a progressive enhancement in cognitive performance (Tramonti Fantozzi et al., 2023). A predominant trigeminal sensorimotor activity on one side may therefore lead to an asymmetry in LC discharge, which can be further enhanced by the reciprocal LC connection between the two sides (Jones and Yang, 1985), which are inhibitory in nature (Cedarbaum and Aghajanian, 1977).

The condition could be detrimental for higher level sensory-motor and cognitive processes (De Cicco et al., 2014, 2016; Tramonti Fantozzi et al., 2019, 2021a,c), either by inducing an asymmetric body representation, or else an asymmetric hemispheric excitability, akin to the phenomenon described by Lomber and Payne (1996). The scanning dynamic of the text can be affected by such an asymmetry, contributing to the increased number of fixations (saccades) observed in the habitual occlusion condition compared to the occlusion corrected condition.

In conclusion, the influence of occlusion condition on oculomotor behaviour can be attributed to both the asymmetric trigeminal input to oculomotor structures and the effects of trigeminal asymmetries on higher visuo-spatial functions.