Earlier experiments have shown that trigeminal signals play a particularly important role in the control of cortical desynchronization and arousal (Roger et al., 1956). This can be accounted for by the influence exerted by trigeminal afferents on the Ascending Reticular Activating System (ARAS). Trigeminal primary and/or secondary afferents reach the pontomedullary reticular formation (McKinley and Magoun, 1942), the noradrenergic neurons of the locus coeruleus (LC) (Cedarbaum and Aghajanian, 1978; Luo et al., 1991; Craig, 1992; Couto et al., 2006), the cholinergic neurons located within the pedunculopontine and the laterodorsal tegmental nucleus (Semba and Fibiger, 1992), the tuberomammillary hypothalamic histaminergic neurons (Fujise et al., 1998; Sakata et al., 2003) and the thalamic non-specific intralaminar and midline nuclei (Krout et al., 2002; Clascá et al., 2012). Although it has been claimed that mesencephalic trigeminal neurons (see Wang and May, 2008) are electrotonically coupled to LC neurons (Matsuo et al., 2015), there is at present no evidence of this statement, beyond a Fluorogold labeling passage from mesencephalic trigeminal neurons to the LC area (Fujita et al., 2012), which only implies the possibility of such a speculation. LC neurons modulate arousal (Samuels and Szabadi, 2008; Carter et al., 2010), sensorimotor excitability (Matsutani et al., 2000) and provide an essential contribution to arousal-associated (Bradshaw, 1967) pupil mydriasis (Gabay et al., 2011), which reflects “mental efforts” (Hess and Polt, 1964) and task performance (Rajkowski et al., 1993). In fact, LC controls the preganglionic parasympathetic neurons of the Edinger–Westphal nucleus (Breen et al., 1983), which innervate the iris constrictor, inhibiting their discharge through an α2-mediated mechanism (Szabadi and Bradshaw, 1996; Samuels and Szabadi, 2008). Such inhibition is necessary to increase the pupil size, since the tonic activity of the iris constrictor would prevent pupil enlargement by dilatator iris (Wilhelm et al., 2011). As a consequence, the LC neurons activity is strongly and positively correlated with the pupil size, both in animals (Rajkowski et al., 1993, 1994; Joshi et al., 2016) and humans (Alnaes et al., 2014; Murphy et al., 2014). For this reason, several studies have utilized the pupil size as an index of LC activity (Silvetti et al., 2013; Hoffing and Seitz, 2015; Kihara et al., 2015; Hayes and Petrov, 2016; see also Laeng et al., 2013). The connection of the trigeminal system with the ARAS and LC suggests that the modifications of trigeminal input occurring during chewing may induce relevant changes in the whole brain, leading to an enhancement in the arousal/alertness level and, as a consequence, in the cognitive performance (Sakamoto et al., 2009).

Mastication is driven by a central pattern generator (Dellow and Lund, 1971), controlled by oral sensory feedback (Appenteng et al., 1982; Lund, 1991), according to the changes in the consistency and texture of the food bolus (Horio and Kawamura, 1989). Masticatory muscle activation (Peyron et al., 2002), together with feedback signals from periodontal receptors and muscle spindles (Lavigne et al., 1987), increases with food hardness and influences brain function. In fact, animals submitted to long-term soft-diet feeding undergo a decrease in learning and memory (Tsutsui et al., 2007; Weijenberg et al., 2013), whereas mastication prevents the degradation of brain functions (Gatz et al., 2006; Okamoto et al., 2010; Ohkubo et al., 2013). Moreover, animal models showed that bilateral molar extractions, leading to long-term masticatory dysfunction, decrease the number of pyramidal cells in the hippocampal CA1 and gyrus dentatus (Oue et al., 2013), with impairment of spatial learning and memory in water maze tests (Kato et al., 1997). These deficits increase with aging and time after teeth loss (Onozuka et al., 2000). On top, teeth loss increases the proliferation and the hypertrophy of the astrocytes within the hippocampus, as it occurs following neuronal degeneration and senescence processes (Onozuka et al., 2000), while decreasing c-Fos expression during spatial task (Watanabe et al., 2002), dendritic spines density (Kubo et al., 2005) and neurogenesis (Aoki et al., 2010).

In humans, beyond the trophic, long-term effects that chewing exerts on the brain (Gatz et al., 2006; Okamoto et al., 2010), gum chewing improves cognitive processing speed (Hirano et al., 2013), alertness (Allen and Smith, 2012; Johnson et al., 2012), attention (Tucha et al., 2004), memory and learning (Allen et al., 2008; Smith, 2009); it also reduces reaction times (Hirano and Onozuka, 2014) and event-related potentials latencies in an auditory oddball paradigm (Sakamoto et al., 2009). Moreover, shortening of the visual reaction time in a button press task following chewing (Hirano and Onozuka, 2014) was found associated with an increase of the blood-oxygen-level dependent signal in the anterior cingulate, left frontal gyrus and motor related regions. On the other hand, chewing-induced improvement of short-term memory processing was coupled to an increased blood flow in the middle frontal gyrus of the dorsolateral prefrontal cortex, the right premotor cortex, precuneus, thalamus, hippocampus and inferior parietal lobe (Hirano et al., 2008).

Chewing-induced advantages in cognitive performance were observed following, but not during chewing bouts for a time period of 15–20 min (Onyper et al., 2011). The consistency of the chewed gum pellet influences chewing-induced performance changes. In fact, at variance with an ordinary chewing gum, a soft chewing gum like bubble gum did not improve memory (Davidson, 2011). On the other hand, the prolonged chewing of a hard gum significantly increased the fatigue of the masticatory muscles, blunting cognitive performance (Farella et al., 2001).

Chewing effects on cognitive performance and arousal could be related to trigeminal action on different ARAS components. More specifically, the involvement of the LC should result in an increased task-related mydriasis, due to the coupling between LC activity and pupil dilatation, well documented by several investigations (Rajkowski et al., 1993, 1994; Alnaes et al., 2014; Murphy et al., 2014; Joshi et al., 2016). Therefore, based on the assumption of the proposed LC contribution to chewing effects on the brain, a correlation can be expected between chewing-induced changes in performance and in task-related mydriasis. Thus, the aim of the present study was to investigate whether short bouts of masticatory chewing influence (1) the velocity of retrieval of specific target numbers within numerical matrices and relative errors as well as (2) the mydriasis induced by a haptic task. Finally, we tested whether chewing induced changes in cognitive performance and mydriasis are correlated with each other.

The present findings indicate that short bouts of masticatory activity improve the velocity of number retrieval and matrix scanning and increase the task-induced mydriasis, which is considered as an indicator of arousal (Bradshaw, 1967; Bradley et al., 2008). It has to be noticed that, when a hard pellet had been chewed, the improvement in performance and the increase in mydriasis, observed immediately after mastication, persisted for at least 30 min. These findings could not be attributed to a learning effect, since simple test repetition did not significantly change sensorimotor performance and mydriasis, neither to other sensory properties as the hard pellet was tasteless, odorless and colorless. Moreover, it was specifically attributable to chewing and not to the motor activation in itself, since handgrip exercise induced slight but significant decreases in performance and mydriasis. Such decreases were significant only among females, likely owing to less developed musculature which may have been fatigued by the handgrip exercise, thus reducing their performance (White et al., 2013). To weigh the possible translational impact, the chewing effects have been replicated by administering a commercially available gum. The results show how the effects can be still observed, but they are short-lasting and reduced in amplitude. We argue that this might be related to the soft consistency of the commercially available pellets, as compared to the custom-made hard ones.

The improvement in cognitive performance induced by chewing could be attributed to an enhancement of the arousal level, as indicated by the task-related enhancement in mydriasis observed following the chewing period, and by the significant correlation between the changes in performance and mydriasis. Thus, these findings suggest that the sensorimotor orofacial activity has a strong impact on the modulatory system regulating arousal, likely due to the influence exerted on the ARAS and related structures by the central pattern generator for chewing and/or by the sensory signals elicited by masticatory movements. In fact, trigeminal input exerts a tonic action on the cortical activity, as indicated by the cortical synchronization induced by bilateral trigeminal neurectomy in encephale-isolé preparation (Roger et al., 1956) and by the cortical desynchronization associated with the stimulation of the trigeminal nerve in awake epileptic rats (Fanselow et al., 2000). Although the fibers arising from the trigeminal nuclei may affect several structures belonging to the ARAS (McKinley and Magoun, 1942; Semba and Fibiger, 1992; Fujise et al., 1998; Krout et al., 2002; Sakata et al., 2003), the present data suggest that, at least in part, the stimulating effects of trigeminal information can be mediated by the LC. In fact, pupil size is widely considered as a proxy of LC activity (Silvetti et al., 2013; Hoffing and Seitz, 2015; Kihara et al., 2015; Hayes and Petrov, 2016; see also Laeng et al., 2013). In the present experiments, the larger changes in task-induced mydriasis corresponded to 30% of the control value (11% of the pupil size at rest) and corresponded to 0.43 mm. In humans, changes in pupil size of 0.25–0.5 mm have been shown to correlate with the corresponding changes in the LC region activity (Alnaes et al., 2014). Within this framework, our experiments document the novel finding of a significant correlation between changes in task-induced mydriasis and performance. Trigeminal inputs to the LC have been described from the spinal and principal sensory trigeminal nuclei (Craig, 1992; Couto et al., 2006) as well as from within or near the trigeminal mesencephalic nucleus (Cedarbaum and Aghajanian, 1978; Luo et al., 1991). It has been claimed that the latter could be electrically coupled to the LC (Matsuo et al., 2015). The modulatory action of the LC on a wide range of trophic brain functions (Mather and Harley, 2016) and the degenerative effects on the brain induced by teeth loss (Kato et al., 1997; Onozuka et al., 2000; Watanabe et al., 2002; Kubo et al., 2005; Aoki et al., 2010; Oue et al., 2013) could be consistent with the speculated coupling. However, there is still no direct evidence of electrocoupling between LC neurons and trigeminal mesencephalic afferents, beyond a Fluorogold labeling passage from mesencephalic trigeminal neurons to the LC area (Fujita et al., 2012). Fibers from trigeminal nuclei could affect the distal dendrites of the pericoerulear region (Aston-Jones et al., 1986, 1991), as well as, via the paragigantocellularis and the prepositus hypoglossi nuclei (Lovick, 1986; Buisseret-Delmas et al., 1999), the LC cell bodies (Aston-Jones et al., 1986). So, an increase in LC activity could occur during the trigeminal-induced enhancements in task-related mydriasis and in cognitive performance (see De Cicco et al., 2014, 2016), which would benefit of the release of noradrenaline at cerebral cortical level (Gabay et al., 2011).

It is unknown whether trigemino-coeruleus pathways modulate the activity of noradrenergic neurons projecting to the thalamus and the cortex, which are likely implicated in the arousal. Recent evidence indicates, in fact, that the LC is heterogeneous in terms of its neurochemical composition, neuronal firing patterns, intranuclear topography of target-specific projection neurons (Uematsu et al., 2015), and sub-regional extension of peri-coeruleus dendrites (Aston-Jones et al., 1991). Further investigations are necessary in order to clarify the issue. However, cells projecting to different brain region are largely overlapping (Uematsu et al., 2015), so that trigeminal afferent input to a given LC region could potentially affect different output channels.

So, it is likely that sensorimotor trigeminal signals elicit changes in the excitability of ARAS/LC which outlast mastication for at least a few minutes and may extend up to half an hour when hard pellets are chewed.

Another relevant point emerges from the improvement in performance. In fact, this resulted from an increase in the scanning velocity of the matrices without loss in accuracy, since the increase in retrieved numbers was not associated with an increase in errors. Such an increase would have been expected on the ground of the well-known trade-off between speed and precision (Fitts, 1954; Wright and Meyer, 1983; Harris and Wolpert, 2006). These data are consistent with the widespread effect of NE release (McCormick et al., 1991), which increases the signal to noise ratio of cortical and thalamic neurons (Foote et al., 1975; Moxon et al., 2007), thus enhancing sensory coding efficiency (Fazlali et al., 2016) and, possibly, network performance in highly integrated functions (Shea-Brown et al., 2008). Finally, the observation that a higher masticatory load and, as a consequence, a stronger trigeminal input induce more significant effects on task-related mydriasis and performance with respect to a less demanding chewing task with a soft, commercially available pellet is consistent with clinical and experimental animal evidence showing how chewing soft food may lead to cognitive impairments and neurodegenerative processes (Tsutsui et al., 2007; Weijenberg et al., 2013). These data have important clinical implications, paving the way to simple behavioral strategies for preventing and/or slowing the progression of neurodegenerative disorders.

Despite a few limitations (small sample size, lack of any cognitive evaluation of the participants), the present findings support the hypothesis that trigeminal activity associated with chewing activates brain arousal systems, possibly the LC, leading to persistent changes in their excitability likely improving the performance of cognitive tasks.

MPTF performed most of data analysis and gave a very important contribution to data understanding, manuscript preparation and revision. VDC contributed to the experimental design and executed the experiments. The contributions of MPTF and VDC were of the same importance for the work. MB gave an important contribution to data analysis and manuscript preparation and revision. EC, UF, and LB contributed to manuscript preparation and revision. DM contributed to the experimental design, supervised the whole activity and prepared most of the manuscript. All authors read, edited and approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer KS and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.