A total of 176 articles were primarily selected for review. Exclusion of articles published prior to the year 2000, with the exception of articles of significant historical value, provided us with a total of 71 articles of potential significance for review. After a reference search, another seven articles were included in the final review. As a result, a total of 78 articles were included in the following review. In order to better portray our results, we categorized the findings into six subsections.

From the phylogenetic viewpoint, the claustrum first appeared in the form of a distinguishable nucleus in marsupials and primitive primates [12], though the claustrum has been also identified in reptiles (as a sleep-regulatory unit) [36]. Brodmann, Ariëns Kappers, Sonntag and Woollard, Rose et al. believed that the claustrum was a derivative of the insular cortex. This assumption was supported by morphogenetic similarity of these structures [12, 13]. But this was disputed by Landau, Faul and Holmgren, who stated that the claustrum is related to the corpus striatum in origin. This hypothesis is supported by reports of abnormal cerebral cortex development with a completely absent insular cortex in one of the hemispheres, while the claustrum remained intact. Additionally, it has been reported that the claustrum develops prior to the insular cortex. These findings give credibility to statements supporting the subcortical origin of the claustrum [12].

Furthermore, the claustrum is not considered to be part of the cortical plate or corpus striatum. It is rather regarded as an intermediary between these structures. Several valid arguments in favor of this theory have been provided, stating that the claustrum developed as a result of neuroblast accumulation from the pallial matrix (which borders the corpus striatum) [12]. These neuroblasts do not reach the cortical plate during migration and are therefore situated at a certain distance between it and the corpus striatum. Other studies have also presented similar findings regarding the ontogenesis of the claustrum. As of today, the hybrid claustrum ontogenesis hypothesis is considered the most acceptable [12].

Analysis of specific genetic markers in the insular cortex and the claustrum show different genetic profiles, which underlines the separate ontogeny of these structures [61]. Interestingly, despite specific genetic profile differences between the insular cortex and the claustrum, these two areas have the most similarities in genetic expression, compared to the rest of the brain regions [61]. Nonetheless, existing data suggests that the claustrum is an independent ontological structure, characteristic of newer and advanced neural function development. Genetic studies have also presented data supporting the claim that the claustrum and amygdala are homologues of the sauropsid dorsal ventricular ridge, though epigenetic, hodological, morphological, and topographical data did not support this hypothesis [85]. The evaluated genetic links show valuable insight into the ontological formation of the mammalian claustrum.

The claustrum is often referred to as the “wall of the brain”. It is a thin, curved sheet of neurons embedded in white matter, located in both brain hemispheres deep to the neocortex. The claustrum is situated between the insula and putamen, separated by the extreme and external capsules respectively [14]. The frontal part of the claustrum is directed towards the corpus amygdaloideum and is divided by fibrous bundles connected to the anterior commissure and ancyroid bundle.

In humans, the claustrum occupies about 0.25% of the cerebral cortex volume [15]. Its thickness varies from a fraction of a millimeter to several millimeters [16]. Its dimensions are on average 38 mm rostrocaudally and 22 mm dorsoventrally [5, 15]. The inferior border of the claustrum is located at the level of the insular cortex and putamen [17]. The claustrum is asymmetrical, reflecting anisotropy between the hemispheres [17,18,19]. The volume of the claustrum ranges from 744 to 912 mm3. It has been reported that the average volume of the claustrum in men is greater than in women (Fig. 1). Claustrum volume asymmetry between hemispheres varies between 15 and 20%, with an average volume of 829 mm3 in the right hemisphere and 706 mm3 in the left [19].

Claustrum volume discrepancies in the human brain

Morphologically, the claustrum corresponds to the concavity of the insular cortex, and its inner curvature follows the convexity of the putamen. The claustrum is completely surrounded by white matter deep in the brain tissue, which complicates the study of its structure and connections. Exact anatomical boundaries of the claustrum still remain uncertain [18,19,20,21]. Blood supply to the claustrum is maintained through insular perforating vessels, deep and superficial branches of the medial cerebral artery [4, 16].

The claustrum is divided into the dorsal and ventral zones [22]. The dorsal (insular or compact) zone is located above the rhinal sulcus, medial to the insular cortex. The dorsal zone represents a continuous sheet of gray matter which expands inferiorly and superiorly [23, 24]. The ventral zone (referred to as the prepiriform, fragmented, amygdaloid, temporal claustrum or endopiriform nucleus [25]) is located medial to the piriform cortex and includes neural complexes underlying the rhinal sulcus [23, 24, 26]. The ventral zone is divided into the upper and lower sections. The upper ventral section is reported to be connected to the anterior pole of the dorsal zone, which extends to the base of the frontal lobe, and adjoins the prepiriform cortex. The lower ventral section connects to the posterior pole of the dorsal claustrum, which is directed towards the amygdaloid region of the cortex. Here the connection between the ventral claustrum and the amygdala is so close that it becomes difficult to differentiate these structures [24].

The presence of multiple classes of neuronal projections and interneurons are reported to exist within the mammalian claustrum [27, 28]. The anatomical proximity of the claustrum and its surrounding structures, underlines the difficulties in neuronal circuit differentiation. This may be due to mutual functional activity, but also due to the drawbacks of existing neural circuit evaluation methods (fMRI). Determining the connection between the registered signal and the activity of the actual neural substrate (claustrum, insula, or putamen) is difficult [8], since local stimulation may result in activation of nearby structures [29]. Small Region Confound Correction (SRCC) can be used to differentiate the claustrum neural circuits from nearby anatomical structures, due to high accuracy of the method [29]. The relationship between the structural and functional aspects of the claustrum still requires extensive research to develop methods that would eliminate existing limitations. Currently, such methods are being developed, allowing for potentially larger-scale more precise studies of the claustrum [30].

Several neuroanatomical studies showed that the claustrum has an extensive network of connections to subcortical regions, including the hippocampus, thalamus, putamen, and basal nuclei, as well as the temporal, occipital, and sensory lobes [18, 24, 31,32,33; Fig. 2). It remains to be determined if there are any cortical regions that are not connected with the claustrum [18, 19]. As such, the human claustrum is considered to be one of the most densely connected structures in the brain per unit volume [15, 31].

The schematic depiction of claustrum neuronal connections with cortical and subcortical structures

Existing tractography studies identified two main projection tracts: the dorsal and ventral. The dorsal tract is a group of fibers that connects the claustrum to sensory and motor regions of the cortex [24, 34]. The ventral tract connects the claustrum to the auditory and olfactory regions [5]. Other sources differentiate four fiber tracts connecting the claustrum to the cerebral cortex: the anterior, posterior, upper, and lateral tracts [18]. The anterior and posterior tracts connect to the prefrontal cortex and visual analyzer areas. The upper tract connects to sensorimotor regions and the lateral tract connects to the auditory cortex. The medial tract connects to the basal ganglia, specifically to the caudate nucleus, putamen, and globus pallidus [35]. Connections with the basal ganglia, however are debated: several input organization studies did not reveal claustrum connections with basal ganglia [87, 88]. Other animal studies also do not find connections between basal ganglia and claustrum. Additional studies have shown connections between the claustrum and the contralateral cerebral hemisphere [86], which include interhemispheric cortico-claustral fibers and inter-claustrum fibers. Tractography studies performed with diffusion tensor imaging, are not a precise way of assessing anatomical connectivity. As such, the true extent and connectivity of the claustrum network is yet to be shown.

Claustral neural circuits show high affinity to the frontal cortex, including the anterior cingulate gyrus, the prelimbic region and medial prefrontal cortex. The primary sensorimotor regions seem to be poorly connected to the claustrum [29]. Claustrum neural circuits functionally link the anterior cingulate gyrus to the visual and parietal cortex and these projections distinguish the inhibitory circuit of claustrum neurons involved in processing and transmitting of information [11]. The claustrum therefore connects with the frontal cortex, the anterior cingulate gyrus and the secondary visual cortex (B1/B2), the parietal associative cortex and coordinates spatial–temporal control of certain cortical areas [11].

The claustrum has been shown to be connected with the cingulate and prefrontal cortex to foster cognition influence [29]. Cognitive task performance is accompanied by the activation and deactivation of certain cortical regions to confer functional connectivity linked to claustrum circuits [29]. Therefore, the human claustrum is reported to play a significant role in cognition control, specific task control independent of sensory-motor processing. Existing studies show that cortico-claustral projections originate from areas of the frontal cortex including the orbitofrontal, cingulate, and secondary motor cortex. The somatosensory, auditory, and visual regions are minimally connected to the cortical-claustral inputs [20]. Cortical-claustral contralateral projections are reported to be much denser than cortical-claustral ipsilateral projections [30]. The claustrum also receives projections from several subcortical structures including the mediodorsal thalamus, basolateral amygdala, and hippocampus [20]. Evidence of intraclaustral connections exists through an extensive network along the Rostro-caudal axis [36].

Several markers were discovered to be specific to the claustrum, such as the g-protein Gamma-2 subunit (Gng2) and parvalbumin-immunoreactive (PV-IR) neuropil [6]. Gng2 expression and other immunohistochemical staining of parvalbumin opens up new possibilities to delineate underlying neuronal network dynamics of claustrum neural circuits [34].

Seventy-five percent of neurons projecting into the visual cortex were shown to be Type I neurons, and 25% were Type II neurons. Parietal oriented neural circuits included 43% Type I and 57% Type II neurons [10]. Among claustral neurons projecting into the anterior cingulate gyrus, 21% were Type I and 79% were Type II. The executive cortex (anterior cingulate gyrus) receives mostly projections from Type II neurons, while the visual area receives mostly projections from Type I neurons. This data shows claustrum potential in mediation of cortical functions [10].

In contrast to the cortex, the claustrum does not exhibit a layered organization. There is relatively simple and uniform cytoarchitecture within a small number of cell types specific to the claustrum [37]. As such, the claustrum is mostly composed of glutamatergic neurons [38] evenly distributed across the claustrum. The diameter of a glutamatergic neuron body is 15–29 μm and the dendrites of these cells are covered with spinules and do not exhibit a preferred orientation [39]. Their axons protrude from the claustrum, mainly into the cerebral cortex. Glutamatergic neurons are excitatory neurons and include pyramidal, fusiform, or circular neurons. However, the functional subgroups of glutamatergic neurons specific to the claustrum have not been fully identified [40, 41]. GABAergic neurons are less widespread, and account for the remaining 15% of neurons within the claustrum [38]. The diameter of the GABAergic neural body is 10–15 μm [22]. Dendrites of these neurons appear smooth and axons do not extend beyond the body of the claustrum [38].

Golgi lipofuscin granular staining showed the presence of five types of neurons within the claustrum, consisting of a combination of spiny and aspiny nerve cells with divergent morphometric properties [42]. Type I cells are spiny neurons with small and widely distributed chromolipoid granules. Type II cells are large aspiny neurons and contain a large number of pigment granules. Type III cells are large aspiny neurons without pigment granules. Type IV cells small aspiny neurons with pigment deposits. Type V cells are small aspiny neurons without chromolipoid granules [42, 43].

Some claustrum neurons express Vglut2, which is a characteristic of subcortical cells [6]. Claustrum neurons can further be differentiated by parvalbumin expression into PV(−) and PV(+) interneurons. Furthermore certain electrophysiological properties including distribution of spike accommodation, maximum firing rate, capacitance, voltage of the resting membrane, and membrane resistance have shown to differ in different groups of claustrum neural cells [6]. Claustrum neurons show reactivity to cortical signals [11, 44]. Further evaluation of electrophysiological features of different claustrum neurons may help unravel the specifics of pathological changes within the claustrum associated with neurodegeneration [19, 45, 46].

A transcriptome-wide analysis of claustrum neural cell component provides the most specific differentiation of specific cellular profiles. As such, NR4A2, NTNG2, and LXN are thought to be more strongly expressed in the claustrum [61]. Furthermore, the specific expression patterns in the claustrum show enriched genes pertaining to severe intellectual disability, epileptic encephalopathy, intracellular transport, spine development, macroautophagy, smoking addiction. The NR4A2 gene plays a role in dopamine regulation, and has been implicated in depression, drug addiction, Parkinson’s disease, schizophrenia and bipolar disorder [61].

Crick and Koch argued that the claustrum performs final integrative processing of input information, through convergence of several neural circuits connected to the cerebral cortex [47, 48]. The function of the claustrum is manifested when different sensory modalities are perceived altogether [49, 50]. It was found that the claustrum can combine visual, tactile, auditory, and emotional sensations (Fig. 3; 51, 52]. Furthermore, the claustrum plays a key role in neural mechanisms of consciousness [53,54,55].

Multimodal information processing through the claustrum

Experimental studies show that inactivation of claustrum neurons typically undermined the efficiency to execute rewarding behavior due to increased susceptibility of such subjects to distracting factors [56]. Claustrum stimulation was found to cause a disruption in attentive behavior and a presumed loss of consciousness [57]. Existing reports suggest that the claustrum plays a significant role in mediating information processing in focus, prioritization and ignoring unnecessary sensory input [