The striatum is a crucial brain region involved in goal-directed behavior, habit, learning, and value process and is associated with various brain disorders, including Parkinson’s disease, Huntington’s disease, and schizophrenia (Cools et al., 2001; Delong and Wichmann, 2007; Ferrante et al., 1985; Howes and Kapur, 2009; Schmack et al., 2021; Simpson et al., 2010; Vonsattel et al., 1985). To better understand human cognitive behavior and develop effective treatments for brain disorders, it is essential to identify the similarities and differences between rodent and primate brains and assess the applicability of data from the rodent striatum to humans.

Among the striatum regions, the caudate head (CDh) in primates and the dorsomedial striatum (DMS) in rodents, both located in the anterior-dorsal-medial part of the striatum, play important roles in motor control, action selection, decision-making, cognitive flexibility, and reinforcement learning (Brovelli et al., 2011; Grahn et al., 2008; Kim and Hikosaka, 2013; Knowlton and Patterson, 2018; Richfield et al., 1987; Yin, 2005a; 2005b; Yin and Knowlton, 2006). Recent studies have suggested that these regions may be functionally homologous, particularly in their involvement in goal-directed behaviors (Broschard et al., 2023; Kim and Im, 2019; Mestres-Missé et al., 2012). However, controversy remains regarding the homology of the primate caudate tail (CDt) and the rodent tail of the striatum (TS), located in the caudal part of the striatum, due to their different anatomical inputs that may reflect distinct functions (Griggs et al., 2017; Jiang and Kim, 2018; Kim et al., 2014; 2015; Menegas et al., 2018; Seger, 2013; Valjent and Gangarossa, 2021; Yamamoto et al., 2013). Therefore, in this review, we examined and compared the functions and anatomical connections between the TS in rodents and the CDt in primates to help clarify their similarities and differences.

One easily recognizable difference between rodent and primate brains is the complexity of the cortical areas. Primate brains have wider cortical areas and more gyri and sulci compared with rodent brains. This difference arises from the developmental process along the rostral-caudal axis of the brain (Fig. 1). After formation of the neural tube, cells proliferate inside the neural tube, and the shape of the tube expands differently across the rostral-caudal axis in rodent and primate brains (Copp et al., 2003; Stiles and Jernigan, 2010). The neural tube of the primate brain stretches more caudally compared with that of the rodent brain, resulting in more prominent structures, such as the temporal cortex and the cingulate cortex (Fig. 1).

In addition to the difference in the surface of the brain, there are also differences in the subcortical structures along the rostral-caudal axis. For example, the location of the hippocampus is different in rodent and primate brains: the hippocampus is in the dorsal part of the rodent brain, while in primates it is in the ventral part of the brain (Figs. 1 and 2) (Clark and Squirea, 2013; Strange et al., 2014). Even more striking than the differences in location are the differences in the shape of the structure arising from expansion along the rostral-caudal axis. The shape of the rodent striatum, which may include the caudate and putamen in primates, differs from that of the caudate in primates (Figs. 1 and 2). Notably, the caudal part of the caudate nucleus extends along the rostral-caudal axis of the primate brain to generate the CDt, which appears different from the rodent TS in the lateral view (Fig. 1) (Griggs et al., 2017; Jiang and Kim, 2018; Kim et al., 2014).

Overall, the expansion of the primate brain along the rostral-caudal axis leads to differences in the location and shape of brain structures compared with those in the rodent brain. Considering the differences in the anatomical circuits of rodent and primate brains along the rostral-caudal axis is important, as these differences may lead to differences in information processing and final output behavior.

The striatum is a nucleus in the subcortical basal ganglia that receives inputs from various cortical and subcortical areas and plays a crucial role in functions such as motivation, goal-directed behavior, and habit (Balleine and O’Doherty, 2010; Liljeholm and O’Doherty, 2012; Tricomi et al., 2009). Although the rodent and primate striatum have been regarded as analogous structures, there are differences in their shape, location, and anatomical connections (Balleine and O’Doherty, 2010; Heilbronner et al., 2016; Woolley et al., 2013). The primate striatum is divided into three regions: the caudate nucleus, putamen, and ventral striatum (Figs. 2A and 2D). In contrast, the rodent striatum, known as the caudoputamen, consists of a single structure in which the caudate nucleus and putamen are not separated (Figs. 2B and 2F). In both rodents and primates, the rostral part of the striatum is the largest, and it tapers as it progresses towards the caudal end, forming a tip referred to as CDt in primates (Figs. 2A and 2C) and TS in rodents (Figs. 2B and 2E). However, the primate CDt is a distinct anatomical region that arises from the caudate nucleus, whereas in rodents, the TS is a structure that may merge with the caudate and putamen. Due to this difference in the striatal organization, whether the tail of the rodent caudoputamen (TS) is homologous to the CDt remains unclear. The differences in the anatomy of the striatum between primates and rodents raise questions about whether these regions have homologous anatomical connections and functions.

Previous studies have shown the cortical and subcortical connections with the TS and CDt (Griggs et al., 2017; Jiang and Kim, 2018). Jiang and Kim (2018) investigated the rostral-caudal distributions of retrogradely labeled TS-projecting neurons in the whole rodent brain and found that the TS primarily received inputs from the caudal regions of the cortex and subcortex, while the DMS received inputs mainly from the rostral regions. Similarly, in primates, CDh-projecting neurons were predominantly located in the rostral regions of the cortex (A17 to A32), whereas CDt-projecting neurons were primarily located in relatively posterior regions (A3 to A11) (Griggs et al., 2017; Haber et al., 2006; Middleton and Strick, 1996; Seger, 2013). Moreover, as retrograde tracer injections progressed more posteriorly into the caudate, the labeled cortical regions topographically shifted posteriorly (Saint‐Cyr et al., 1990). This suggests that the distribution of striatum-projecting neurons is driven by the rostral-caudal axis rather than being heterogeneously distributed.

The subcortical inputs into the TS and CDt have been identified. Griggs et al. (2017) demonstrated that neurons in the caudal claustrum, thalamus, lateral substantia nigra pars compacta (SNc), and medial dorsal raphe nucleus (DRN) innervated the CDt in the macaque monkey (Fig. 3A) (Griggs et al., 2017; Kim et al., 2014). Similarly, in rats, the TS receives inputs from the claustrum, thalamus, lateral part of the SNc and the DRN (Fig. 3B) (Jiang and Kim, 2018).

Outputs of the primate CDt were identified by anterograde tracer injections. Labeled axon terminals of medium spiny neurons located in the globus pallidus (GP) and substantia nigra pars reticulata (SNr), which are the motor output structures in the basal ganglia (Fig. 3A) (Hikosaka and Wurtz, 1983; Parent et al., 2000; Saint‐Cyr et al., 1990; Zhou and Lee, 2011). The terminals of the rodent TS are also distributed to the GP and the lateral edge of the SNr (Fig. 3B) (Haber et al., 2000; Ogata et al., 2022; Tulloch et al., 1978). This suggests that the TS and CDt receive value inputs from dopamine neurons in the SNc and serotonin neurons in the DRN and send motor outputs through the SNr. This anatomical evidence suggests that the rodent TS and primate CDt may have similar functions in reward value-guided motor control. Therefore, the TS and CDt can be considered homologous structures based on their anatomical positions and subcorticostriatal connections (Fig. 3).

There are differences in the anatomical connections between the rodent TS and the primate CDt when considering the inputs from the sensory cortex and thalamus. In the primate brain, the CDt is innervated by neurons in the ventral inferior temporal cortex, which mainly processes visual information (Fig. 3A) (Baizer et al., 1993; Griggs et al., 2017; Saint‐Cyr et al., 1990). CDt-projecting neurons were found in the medial dorsal nucleus (MDN), suprageniculate nucleus (SGN), and lateral geniculate nucleus (LGN) of the thalamus, which are responsible for monitoring saccadic eye movements, processing visual value, and relaying visual information, respectively (Fig. 3A) (Griggs et al., 2017; Hu and Jayaraman, 1986; Kim et al., 2020; Sommer and Wurtz, 2008; Takada et al., 1985). Thus, the primate CDt appears to receive inputs from cortical and thalamic regions specialized in visual processing.

In contrast to the primate CDt, the rodent TS receives inputs from various sensory modalities, including the visual, auditory, somatosensory, temporal, and endopiriform cortices, suggesting its involvement in processing multiple sensory modalities (Fig. 3B) (Jiang and Kim, 2018). Moreover, the TS receives inputs from sensory regions in the thalamus, including the lateral geniculate nucleus, medial geniculate nucleus, and ventral posterior nucleus, which process visual, auditory, and somatosensory information, respectively (Fig. 3B) (Guillery and Sherman, 2002; Jiang and Kim, 2018; Landisman and Connors, 2007; Ogata et al., 2022; Oliveira-Maia et al., 2011). These findings suggest that the primate CDt is specialized for processing visual information, while the rodent TS is involved in multisensory processing.

The CDt is involved in multiple aspects of visual processing such as object recognition, visual salience, and object value-based habitual behavior (Brown et al., 1995; Caan et al., 1984; Fernandez-Ruiz et al., 2001; Kim and Hikosaka, 2013). Neurons in the CDt process both feature and location information of visual objects to guide saccadic eye movements (Yamamoto et al., 2012). Moreover, several studies have demonstrated that the CDt and its connected structures are involved in visual habit, where primates automatically make saccades towards valuable objects (Hwang et al., 2022; Kang et al., 2021; Kim, 2021; Kim and Hikosaka, 2013; Mishkin et al., 1984; Yamamoto et al., 2013). Neurons in the CDt selectively encoded long-term value memory of visual objects, and inactivation of CDt neurons impaired habitual saccade to previously learned high-valued objects (Kim and Hikosaka, 2013). Taken together, these electrophysiology and behavior studies with the primate brain indicate that the CDt contributes to encoding long-term value memory of visual objects and guiding automatic saccades toward valuable objects (Kim and Hikosaka, 2013; Yamamoto et al., 2013).

Recent studies have shown that in rodents, the TS processes sensory signals from multiple sources beyond visual stimuli. For instance, optogenetic activation of TS neurons in rats resulted in a bias against ipsilateral choices in auditory decisions, while inactivation of these neurons weakened auditory discrimination performance (Guo et al., 2018). Additionally, dopamine axon terminals in the TS responded to stimuli across different modalities, including visual, auditory, olfactory, and somatosensory, with a higher response to novel and intense stimuli, implying the role of the TS in the processing of the novelty and intensity of stimuli (Akiti et al., 2022; Menegas et al., 2018).

The caudoventral portion of the putamen is located adjacent to the CDt and can be anatomically regarded as the tail of the putamen (PUTt) (Fig. 3A). PUTt and CDt share anatomical inputs and outputs, and both regions receive inputs from extrastriate and temporal visual cortical areas that are responsible for object recognition in the ventral visual pathway (Kravitz et al., 2013; Yeterian and Pandya, 1995; 1998). The GP and SNr are the output structures of the PUTt and CDt, which are connected to the superior colliculus and motor cortex. Like the CDt, the PUTt is also innervated by dopamine neurons in the SNc.

From previous data that the PUTt shares similar anatomical connections with the CDt, it can be assumed that the PUTt may also play a role in habit formation and expression (Saint‐Cyr et al., 1990). Indeed, previous research has shown that the PUTt is involved in habitual behaviors. Buerger et al. (1974) found that lesions in the caudoventral putamen including the PUTt impaired retention of visual discrimination but did not affect their ability to perform auditory discrimination tasks. Fernandez-Ruiz et al. (2001) also demonstrated that neurotoxic lesions in the area including the PUTt impaired learning of 24-h intervals between learning trials for a given stimulus pair but did not affect performance in a delayed non-matching to sample tasks. Finally, a recent electrophysiology study identified neurons in the PUTt that encode the long-term value memory of visual objects and possibly control habitual behavior (Kunimatsu et al., 2019). Previous studies, along with the anatomical similarities between the PUTt and CDt, suggest the involvement of the PUTt in visual habit. However, whether the two brain regions contribute differently to habitual behavior and how they work together or separately to achieve the same goal remain unclear. Further research is needed to address these questions.

Previous studies have shown differences in the anatomical inputs to the rodent TS and the primate CDt. Notably, brain regions processing all five sensory modalities project to the rat TS, while visual areas selectively project to the macaque CDt (Fig. 4A) (Griggs et al., 2017; Jiang and Kim, 2018; Kim et al., 2014). This anatomical difference suggests a possible difference in the way sensory information is processed: a modality-converged process in the rodent striatum versus a modality-selective process in the primate striatum (Fig. 4A). Anatomy and electrophysiology results indicate that the caudal striatum is mainly involved in the generation of automatic behaviors based on long-term memory (Brown et al., 1995; Choi et al., 2020; Guo et al., 2018; Jiang and Kim, 2018; Kim, 2021; Kim et al., 2014; Kim and Hikosaka, 2013). Along with the previous results, we can expect differences in the sensory-to-action outcome process through the caudal striatum between rodent and primate brains.

First, it is possible that neural pathways from five sensory inputs in the rodent brain are integrated into single neurons within the TS (information integration model) (Fig. 4B). In this case, inputs from different sensory modalities may not be fully discriminated, resulting in a non-selective motor response that is not associated with a particular sensory modality. For example, a visual stimulus associated with a reward produces automatic nose-poking behavior, but an auditory stimulus (or any other salient sensory stimuli) may also produce the same nose-poking behavior (cross-modality confusion) (Fig. 4B).

Another possibility is that separate groups of neurons within the TS selectively receive each of the five sensory inputs (information divergence model) (Fig. 4C, left panel). In this case, however, the TS may have fewer modality-selective neurons compared with the CDt, which has neurons that selectively respond to visual stimuli in the primate brain (Fig. 4C, right panel). Having a greater number of neurons dedicated to visual processing may increase the ability to process a wider range of visual stimuli and generate the associated motor outcomes. There may not be enough neurons in the rodent TS to process a large number of visual stimuli compared with those in the primate CDt, resulting in a limit on the capacity for visual stimuli–action associations (within-modality confusion) (Fig. 4C, left panel).

Thus, the possibility of the modality-selective system in the primate striatum (Fig. 4A) may lead to a greater capacity to encode a wider range of sensory stimuli–action associations and better discriminate between sensory stimuli to generate automatic responses more accurately according to various conditions compared with the modality-converged system in the rodent striatum (Fig. 4C, right panel).

It is worth noting that a single region in the rodent striatum receives input from all five sensory modalities, making it easier to find significant results in a single structure study using any type of sensory modality (Akiti et al., 2022; Guo et al., 2018). However, to apply knowledge obtained from rodent studies to primates, it is necessary to identify the regions in the primate brain that process each sensory modality and decode their neural codes.

Previous studies on the anatomy of the primate brain have shown that different regions of the striatum receive inputs from different cortical areas that selectively process each sensory modality. For example, the insular cortex, auditory cortex, and somatosensory cortex project to subregions of the putamen, but not to the CDt (Flaherty and Graybiel, 1991; Fudge et al., 2005; Griggs et al., 2017). These findings suggest that each subregion of the primate caudate and putamen may process the value information associated with each modality. However, how value information of each modality is processed to generate each sensory modality-induced automatic action in the primate brain remains unclear. Addressing this question will be crucial for understanding how primates segregate and combine values associated with diverse sensory modalities.

In this review, we highlight the importance of considering the similarities and differences in the anatomy and function of the striatum between rodent and primate brains. Understanding the similarities and differences can provide insights into how the human brain processes and integrates different types of sensory information and how this varies across different animal species. This knowledge can further our understanding of the neural basis of primate perception, cognition, and behavior and may have implications for the study and treatment of brain disorders.

This work was supported by the Neurological Disorder Research Program (NRF-2020M3E5D9079908), the Korea Research Institute of Bioscience and Biotechnology (KRIBB) Research Initiative Program (KGM4562121) and the Basic Science Research Program (NRF-2019R1A2C2005213) of the National Research Foundation (NRF) by the Korean government (MSIT). Creative-Pioneering Researchers Program through Seoul National University supported this work.

K.L., S.A., J.P., S.L., and H.F.K. wrote the original draft. K.L. and H.F.K. reviewed and finalized the manuscript.

The authors have no potential conflicts of interest to disclose.