Overall, fish and mammals have similar brain topologies. The fish forebrain pushes forward during development rather than wrapping itself around the lower brain regions as it does in mammals ('eversion' rather than 'evagination'). The fish pallium is nevertheless homologous to the mammalian cortex, with distinct sensory and motor regions, although in fish a disproportionate amount of visual processing takes place in the tectum, a homologue of the mammalian superior colliculus (Salas et al., 2003). Likewise, the subpallium of fish corresponds to the mammalian basal nuclei, including the striatum, the key recipient of dopaminergic reward in the mammalian brain. So far so good; most forebrain structures involved in reward processing appear to be conserved among vertebrates, and cognitive abilities previously thought of as exclusive to 'higher' animals (i.e. birds and mammals) are now being studied also in fish (Salas et al., 2003).
However, the fish dopamine supply is all over the place, quite literally. Dopamine neurons are found throughout the zebrafish brain, except for the midbrain, where almost all mammalian dopamine neurons are located. A few dopaminergic clusters in the hypothalamic region project to the subpallium/striatum and were previously thought to be homologous to the mammalian mesolimbic dopamine system, but more recent research has debunked this view (Schweitzer et al., 2011; Tay et al., 2011). There simply is no mesencephalic dopamine system in the zebrafish brain. Nevertheless, fish are capable of both classical and operant reward conditionning (Valente et al., 2011), including dopamine-dependent place preference, and even intracranial self-stimulation (Boyd & Gardner, 1962), so what gives?
(Figure adapted from the Zebrafish Brain Atlas)
As far as I can tell, dopaminergic reward mechanisms remain remarkably poorly understood in the zebrafish, despite intense research in recent years on the neurobiology and genetics of this model system. Most of the dopamine in the zebrafish subpallium/striatum appears to originate in local dopaminergic projections from neurons whose cell bodies are distributed throughout the subpallium/striatum (Tay et al., 2011). These neurons look like plausible mediators of reward, but their input, physiology and function remains unknown(!). The function of the ascending dopaminergic fibres that project to the subpallium/striatum is also not known. Moreover, pretectal dopamine neurons arborize extensively in the tectum, suggesting a possible role in visually guided reward-seeking behaviour, such as hunting.
Plenty of reward-related research to be done in other words, but what do we make of this? Hills (2006) argues that the evolution from anamniotes (fish and amphibians) to amniotes (reptiles, birds and mammals) involved a number of changes regarding dopamine and reward-processing, including:
- The number of cortical imputs to the striatum increased significantly
- The number of dopaminergic inputs to the striatum increased significantly
- The synaptic machinery that allows dopamine to modulate cortical input to the striatum expanded to include DARPP-32
- The dopaminergic signal transitioned from representing the presence of food to representing the expectation of reward more generally
As a consequence of these changes, Hills argues, amniotes were able to apply the neural mechanisms of foraging (e.g. 'area-restricted search', the ancestral function of dopamine, present even in worms and mollusks (Barron et al., 2010)) to search for any kind of information or goal, whether internal or external to the brain; a profoundly powerful adaptation. In addition to the four changes suggested by Hills I would add:
- Dopaminergic cell clusters became centralized in the midbrain
This centralization, together with some specific adaptations, such as gap junctions connecting dopaminergic axons, allowed amniote brains to generate a single, scalar reward signal that adjusts dopamine concentrations homogenously throughout the forebrain.
I think what I need to ask now is: how does the more ancient dopamine reward system of fish actually work; what forms of reward-processing is it capable of; and does its distributed anatomy offer any advantages to the animal or to attempts to understand the neural basis of reward-based cognition?