As discussed, parallel processing between the neocortical and subcortical networks is central to the efficient transfer of information via the brain for both sensorimotor behavior as well as language functions (Hasanbegović 2024; Logothetis et al. 2010; Rutledge and Doty 1962; Ojemann 1991; Schiller and Tehovnik 2015), which is a challenge to the serial models often embraced by neuroscientists who have focused exclusively on the neocortex and its trans-cortical pathways (e.g., Fellman and Van Essen 1991; Mishkin et al. 1983; Wang et al. 2012). Recent advances in recording technology have established, beyond a doubt, that the central nervous system, as a rule, functions according to parallel operations, as illustrated using behaving mice.

Khilkervich, Mrsic-Flogel et al. (2024) studied the brain-wide transformations of sensory information into the execution of a motor act in adult behaving mice using a temporal-frequency visual detection task. A mouse with the head-fixed was positioned on a track wheel facing two screens, each centered on one eye; gratings were presented on each screen and drifted downward at a common temporal frequency (for methods see: Orsolic et al. 2021). After a delay (which was variable and ranged from 3-16 seconds), the temporal frequency of the gratings was abruptly increased, and a mouse was given 2 seconds to lick a spout for a liquid reward on 70% of trials. Licking the spout outside the temporal increase was punished by a denial of reward. On the remaining 30% of trials (catch trials) no licking response was required. After training, the accuracy of licking at the correct time surpassed 80% correctness, and the latency to lick reached an asymptote of ~ 400 ms[1]. During the presentation of the gratings, a mouse was required to remain immobile on the track wheel up until the change in temporal frequency of the stimulus, otherwise the trial was cancelled.

Using silicon-probe electrodes positioned throughout the central nervous system, 12,772 stable units located in the neocortex, the hippocampus, the basal ganglia, the thalamus, the cerebellum, and the brain stem of 15 mice trained on the temporal-frequency visual detection task were studied (Khilkervich, Mrsic-Flogel et al. 2024). It was found that neurons anticipating the elevation in the temporal-frequency of the visual stimulus during the delay period were widely distributed throughout the central nervous system: up to 45% of the neurons were modulated in the visual thalamus, the visual cortex (including primary and secondary areas), and the superior colliculus (superficial layers), and up to 25% of the neurons were modulated in the anterior cingulate cortex, the anteromedial cortex, the prefrontal cortex, the orbital cortex, the primary motor cortex, the posterior parietal cortex, the hippocampal formation (dentate gyrus, CA3, CA1, and subiculum), the striatum, the substantia nigra (reticulata), the superior colliculus (deep layers), the pretectum, the cerebellar lobules (4, 5, 6, Crus I, Crus II, and the vermis), the cerebellar nuclei, and the midbrain reticular formation. Also, the neural responses in the visual areas occurred much earlier than the neural responses beyond the visual areas. There was an absence of response in the orofacial motor nuclei of the medulla that control drinking, since the drinking musculature was suppressed during the delay period. Finally, during the execution of a licking response all the areas examined including the orofacial nuclei in the medulla were activated such that over 40% of the cells discharged per area on average (see Fig. 1l of Khilkervich, Mrsic-Flogel et al. 2024). This recruitment of even the sensory areas during response execution highlights that the sensory areas receive information about the motor state of an animal, which agrees with the extensive anatomical connectivity of the mouse brain as documented by the Allen Institute (Froudarakis et al. 2019).

Accordingly, much of the central nervous system (save the motor nuclei) is being prepared to execute a motor response during a delay period as an animal remains immobile, and the entire brain including the sensory areas are mobilized during response execution. This conclusion is not visuo-centric, since comparable results occur when using a tactile-triggered response-delay paradigm examining neocortical-cerebellar loops in mice (Hasanbegović 2024), although in this series of studies recordings did not capture the neurons beyond the neocortical-cerebellar loops such as in the hippocampus or striatum, for example.

When the parallel pathways between the neocortex and subcortex are completely transected at the level of the pons and the thalamus, consciousness is abolished in humans (Plum and Posner 1980), especially once all the islands of connective tissue from the neocortex have been severed. We have argued previously that when both primary and secondary languages are eliminated through stroke in multilingual speakers such individuals lose all consciousness (i.e., memory) of language. The same would apply to any other faculty represented in the neocortex. Interestingly, when Penfield and Jasper (1954) were removing large parts of the neocortex that did not include the language areas (to interrupt epilepsy), an awake patient continued to communicate during the removal without interruption as though to suggest that consciousness as it pertains to language is blind to such removals[2].

Now, whether the parallel pathways between the neocortex and subcortex of lower mammals subserve consciousness given the reduced number of neurons [e.g., the mouse brain has about 0.09% of neurons of the human brain: 71 million neurons in the mouse and 86 billion neurons in the human, Herculano-Houzel 2009] will be assessed in a future chapter—by arguing that neural design is more important than neural number (Merker 2007), and that the neural number scales with longevity.

Summary

1. The neocortex and its subcortical targets comprise the parallel processing loops that mediate sensorimotor and linguistic transformations.

2. The entire central nervous system is engaged (including the neocortex, the hippocampus, the cerebellum, and the brain stem), as a mouse anticipates a sensorimotor response. This is true for behaviors initiated by vision and touch, and therefore it would be surprising if this property does not also apply to the other exteroceptive senses, such as audition, taste, and olfaction.

3. During response execution by a mouse the entire brain including the sensory areas are activated indicating that the sensory areas are always privy to the motor state of an animal. As we have argued, feedback (whether from internal or external sources) is central to establishing the constitution of an organism.

4. Consciousness (in humans) is abolished if the pathways between the neocortex and subcortex (including the cerebellum) are cut; this negates the functionality of the neural loops that mediate the conscious faculties of the neocortex.

5. Also, consciousness is organized according to islands of neocortical tissue that are interconnected for a common function. If the neurons that code for verbs, for example, are deleted then one’s sentences will become verbless by affecting the entire neocortical-cerebellar loop of neurons (see Ojemann 1991).

Footnote:

[1] Once a mouse is trained on visual detection, the pupil size decreases in anticipation of the change in the temporal frequency of a grating (Fig. 3J of Orsolic et al. 2021) to suggest that pupil size is being diminished to enhance image clarity (Chen et al. 2016; Suryakumar and Allison 2015). This runs contrary to the notion that an increase in pupil size enhances vision (Ghosh and Maunsell 2024), an increase that is often associated with interoceptive/introspective body states (Aston-Jones and Cohen 2005; Pomè et al. 2020).

[2] After removing large regions of neocortex (up to a complete hemispherectomy) while sparing the language areas in alert and behaving epileptic patients, it was found that linguistic consciousness was not affected (the patient kept talking to the surgeon) and the neocortical removals went without notice by the patient (Penfield and Jasper 1954). This type of observation made during brain surgery was evidenced in over 700 operations. Thus, consciousness is modular and function-based, and awareness of cortical insult only occurs when a subject becomes aware of having lost that function, such as losing the ability to think and speak in a language following a stroke (Mariën et al. 2017).