Roman roads represented the core infrastructure of the empire constructed for political and economic unification of the nations. Archeological studies revealed that the roads were constructed with a consistent design while people made the best use of locally available materials. The heavy cost invested in their construction contributed to their exceptional longevity because their design was reasonable. The “construction” of scientific disciplines also ought to be like this.

The science of neuroeconomics, first coined by Glimcher (Reference Glimcher2003), owes a lot to the concept of optimal foraging in behavioral ecology (Stephens & Krebs, Reference Stephens and Krebs1986). The patch use model, particularly the marginal value theorem (MVT) introduced by Charnov (Reference Charnov1976), had a huge impact on this field because of the transparency of its graphical explanation and applicability to a wide range of animals such as great tits (Cowie, Reference Cowie1977), paper wasps (Kasuya, Reference Kasuya1982), starlings (Kacelnik, Reference Kacelnik1984), and humans (as for information foraging) (Pirolli, Reference Pirolli2007). System neuroscientists have designed several behavioral tasks intended to reflect the ecological considerations in their neurophysiological studies.

MVT itself has left room for further sophistication of its mathematical aspects, while system neuroscientists have produced a flood of data ranging from single unit neurophysiology to functional magnetic resonance imaging (fMRI). Unfortunately, the data on the behavior of neurons often lose its relevance to ecological reality. We, therefore, need a good set of both mathematical and ecological theories. In this concise but dense book, the authors revive the classical MVT by adopting canonical mathematical analysis. The authors are also successful in digesting ample neurophysiological data on their mathematical platform, and they formulate clear links between foraging behavior and neuroscience.

In this commentary, we would like to draw attention to several issues that remain unaddressed in this book of “Vigor.” In chapter 4, the authors focus on superior colliculus (SC), and introduce readers to the two competing systems of SC, namely, one for holding still (fixation) and another for moving (saccade to targets). They construct a clear model of “build-up to threshold” neurons that account for the reaction time. Here arise two questions: (1) How and where is the associated utility computed? (2) What dynamics underlie the choice at the level of the SC and downstream? The authors return to the first question in chapters 5 and 6, but they leave the second question partially addressed.

In recent decades, we have witnessed huge advances in our understanding of the representation of utility in cortical structures such as frontal eye field, lateral intraparietal sulcus, orbitofrontal cortex, and dorsal anterior cingulate cortex (dACC). Of particular importance is the study by Hayden, Pearson, and Platt (Reference Hayden, Pearson and Platt2011), which revealed stay-time related neural activities in dACC in a task that mimicked patch-use behavior. The idea of “build-up to threshold for decision making” proved central to understanding the machinery of foraging control. However, a question arises: What determines this threshold? To find the answer, research efforts must be directed toward the sub-SC control in the brainstem.

We may assume that the decision to move occurs in a single step at some higher level of the brain. But we can also consider multiple decision steps distributed along the entire brain/spinal cord network. A study by one of the commentators addresses the control of swimming direction in lamprey spinal cord, which is composed of coupled segmental neural oscillators for locomotor pattern generation. The lamprey instantaneously transitions from forward to backward swimming, and the reversal of the swim direction was reproduced when the caudal segment received stronger excitation than the rostral (Matsushima & Grillner, Reference Matsushima and Grillner1992). A commonly accepted idea is that the brain sends descending signals that change the direction by biased excitation to the caudal spinal cord segments. We may otherwise assume that the brain is informed of the change in swim direction post-hoc, only after the spinal network has decided it via inter-segmental interactions. If this is the case, it is implied that the brain center can only predict the swimming direction with uncertainty.

The possibility of multiple decision steps and the involvement of lower brain mechanisms in decision-making lead us to question the evolutionary origin of optimal behaviors. The authors emphasize the role of the cerebral cortex in the computation of utilities (Ch. 5), which probably applies to us humans and most mammals with developed cortical structures. But what about fishes, amphibians, and birds? Lower vertebrates also have layered cortex (Suryanarayana, Robertson, Wallén, & Grillner, Reference Suryanarayana, Robertson, Wallén and Grillner2017), but these are underdeveloped when compared with mammals. Birds have a big cerebrum comparable to that of primates (Olkowicz et al., Reference Olkowicz, Kocourek, Lučan, Porteš, Fitch, Herculano-Houzel and Němec2016), but it is not layered and the architecture of telencephalic pallia is different from the mammalian cortex (Shanahan, Bingman, Shimizu, Wild, & Güntürkün, Reference Shanahan, Bingman, Shimizu, Wild and Güntürkün2013). Do these non-mammalians put up with sub-optimal decisions based on awkward payoff computation because they lack the mammalian-like cerebral cortex? Certainly, they do not.

The book of “Vigor” concludes by tentatively assigning serotonin to effort (Ch. 7), and two cited papers are particularly important in this argument. The first paper by Lottem et al. (Reference Lottem, Banerjee, Vertechi, Sarra, Lohuis and Mainen2018) examined the effects of optogenetic activation of raphe nuclei (serotonergic cells) on foraging tasks. Contrary to the widely accepted idea that serotonin tames animals (i.e., inhibitory control), the study suggests that serotonin promotes resource exploitation and suppresses exploration of new patches; this is in concert with our pharmacological study using serotonin-selective reuptake inhibitor (Matsunami et al., Reference Matsunami, Ogura, Amita, Izumi, Yoshioka and Matsushima2012). The second paper is by Cohen, Amoroso, and Uchida (Reference Cohen, Amoroso and Uchida2015), who carefully studied activities of serotonergic neurons and found that (1) tonic activities could represent the global capture rate, whereas (2) transient activities the local capture rate.

The evolutionarily conserved lower brainstem could, therefore, be sufficient for behavioral variables necessary for the control of vigor. If so, how do these serotonergic neurons compute the signals? Do cortical networks instruct the lower brainstem? Or do they uniquely compute the capture rate at the level of the lower brainstem and allow the higher center to be informed of it subsequently? Evolutionary considerations teach us that animals usually make the best use of available “construction materials” to adapt to environmental changes, and rarely innovate new machinery to introduce novel behaviors. The stories of dopamine and serotonin are linked to the phylogenic origins of human neuroeconomics of vigor.

The control of vigor in social setting is another challenging topic that needs to be addressed in the future. When competing for limited resources such as food and mates, animals and humans often invest more efforts; this is known as “social facilitation” (Zajonc, Reference Zajonc1965). When foraging together, animals pay more physical effort without an increase in the capture rate (Amita & Matsushima, Reference Amita and Matsushima2011; in domestic chicks; Ogura, Masamoto, & Kameda, Reference Ogura, Masamoto and Kameda2020 in humans; Ogura & Matsushima, Reference Ogura and Matsushima2011), therefore they apparently deviate from optimality. Competition also causes impulsive choices, encouraging animals choose small-but-immediate reward more frequently even though the alternative delayed option is larger (Amita, Kawamori, & Matsushima, Reference Amita, Kawamori and Matsushima2010; also see Ogura, Amita, & Matsushima, Reference Ogura, Amita and Matsushima2018 for an adaptive value of impulsiveness). Notably, the social facilitation of effort investment was not impaired by selective depletion of mesolimbic dopamine (Ogura, Izumi, Yoshioka, & Matsushima, Reference Ogura, Izumi, Yoshioka and Matsushima2015). We may have to consider variables other than the local/global capture rates when animals interact.

The field of neuroeconomics is still “under construction.” To develop this novel science, we require a lot of “locally available materials” (understanding of foraging behaviors in a variety of animals with different evolutionary and ecological backgrounds) and “a consistent design concept” (common mathematical frameworks). In this book, a gap exists between the concept of metabolic optimization of human locomotion and the mathematical formulations for optimal foraging behaviors (Chs. 1 and 2). Another gap is found between the discussion on the ecology of foraging behavior and the eye movement studies in monkeys, the cortical representations of utility, and the aspects of neuromodulation in mice (Chs. 3, 4, and 5). Despite these few uncomfortable gaps, readers will find the proposed design concept highly valuable, inspiring novel approaches for the unification of different but highly related scientific disciplines of ecology, evolution, economics, and neuroscience in the coming decades.