Proefschrift_Holstein

Chapter 8

The SNS account requires a detailed topographically specific organization of the midbrain and its connections to the striatum and thus far it remains unclear whether this organization, which has been shown in non-human primates and rodents (Haber et al., 2000; Bjorklund and Dunnett, 2007; , but see Matsuda et al., 2009), exists in humans. Indeed, rodent work has suggested that the spiralling dopamine connections between ventral- and increasingly dorsal parts of the striatum are organized in a serial manner. This was evidenced recently by the observation that changes in dopamine signalling in the nucleus accumbens shell influenced dopamine levels in the nucleus accumbens core, which in turn affected signalling in the ventrolateral and dorsal striatum in a serial manner (Ikeda et al., 2013). Several decades ago the idea was proposed that the ventral striatum serves as a limbic-motor interface, mediating the interaction between motivation and action (Mogenson et al., 1980). However, at the time it was unclear how the ventral striatum would do this, apart from the theorizing that midbrain dopamine would be involved. A recent theory has suggested a similar role for the ventral striatum (Mannella et al., 2013) and its input from the prefrontal cortex, amygdala and hippocampus, but it still does not clearly address exactly how reward information can be transferred to motor areas. In chapter 1 we propose that this information transfer occurs via cognitive control regions, rather than via direct connections between motivation and action regions. Evidence from human (Draganski et al., 2008; Choi et al., 2012), non-human primate (Haber, 2003) and rodent (Oh et al., 2014) work has not revealed any direct connections between cortical reward-processing areas and the motor striatum or between the ‘reward midbrain’ (VTA) and the ‘motor striatum’(the putamen). In fact, Ikeda and colleagues (2013) showed that information is processed serially, and that it is ‘forwarded’ from more ventromedial regions of the (rodent) striatum to increasingly more dorsolateral regions. Therefore, it is anatomically not plausible that this information transfer occurs via direct CS connections from reward areas to the motor circuit. In chapter 7 we directly assessed the interaction between reward, cognition and action, showing that integration across these corticostriatal circuits occurs. However, determining whether information bypasses the ‘cognitive’ striatum requires a replication of the current finding after inactivation or lesion of this region, which is not feasible in human subjects ( future research ). Neuroimaging work (e.g. Aarts et al. 2010 and chapter 4 ) has revealed a role for the caudate nucleus in motivated cognitive control, while a role for the ventral striatum in motivated cognitive control is shown in chapter 6 (and hypothesized in chapter 5 ). Although these results may seem contradictory at first, they are in fact perfectly in line with our hypothesized underlying neural mechanism. When interpreting the results from chapter 4 , it is important to realize that the BOLD response is thought to reflect the input from other regions, rather than its output to other regions (Logothetis et al., 2001). The neural signal in the caudate nucleus will thus reflect the input from regions projecting to the caudate nucleus. This is perfectly in line with the idea that information in the ‘reward striatum (the rodent nucleus accumbens or human ventral striatum / anterior caudate nucleus) is transferred to the ‘cognitive striatum’ (the rodent dorsomedial striatum or the human (posterior) caudate nucleus). Excessive NMDA receptor activation in the nucleus accumbens ( chapter 6 ) will have caused neuronal death

168