In section 3.1, De Dreu and Gross contrast reward seeking with loss aversion and conflate behavioural inhibition with fear and active avoidance. We argue that this confuses: (1) valuation with motivation; (2) anxiety with fear; and (3) reinforcers with reinforcement. Making these distinctions has consequences for their proposed neuropsychology.

The expectation/availability (innate or learned) of gain elicits approach. However, omission/termination of expected gain elicits defensive withdrawal (Adelman & Maatsch Reference Adelman and Maatsch1956) and attack (Gallup Reference Gallup1965; Kelly & Hake Reference Kelly and Hake1970), as does an explicit aversive stimulus, such as shock (Renfrew & Hutchinson Reference Renfrew, Hutchinson, Satinoff and Teitelbaum1983). Importantly, even in the presence of loss aversion (Kahneman & Tversky Reference Kahneman and Tversky1979; Novemsky & Kahneman Reference Novemsky and Kahneman2005; Tversky & Kahneman Reference Tversky and Kahneman1991), approach tendencies can be stronger than withdrawal (Hall et al. Reference Hall, Chong, McNaughton and Corr2011), likely as a result of their different goal gradients (Brown Reference Brown1948). So, approach/withdrawal motivations are controlled independently of gain/loss valuations (Hall et al. Reference Hall, Chong, McNaughton and Corr2011), and it is important to keep valuation and motivation theoretically separate (Corr & McNaughton Reference Corr and McNaughton2012) and always take into account the role of contingency (Fig. 1).

Figure 1.

Relations between external amount, contingency, and value. An external item will have a specific amount (e.g., one entire cake) that, together with the current level of drive (which acts like a currency exchange rate) for that kind of item for that person, determines its primary internal value (thickness of arrows in first column). As shown for the case of $1, this interacts with whether the item will be gained or lost to determine the direction and size of its internal value, as ultimately measured by the effect on behaviour. The direction of this effect is reversed if the [expected] gain or loss is omitted. Loss (removal from a store of items) is most easily controlled with money but will also occur when, for example, one rat steals the food from another rat. Reprinted, with permission, from McNaughton et al. (Reference McNaughton, DeYoung, Corr, Absher and Cloutier2016).

It is also important to keep “anxiety” separate from “fear.” Despite their frequent semantic conflation (McNaughton Reference McNaughton2018), the neuropsychology and psychometric evidence are clear on their differences (Corr et al. Reference Corr, DeYoung and McNaughton2013). In contrast to a fear/withdrawal system that is sensitive to threat, the anxiolytic-sensitive Behavioural Inhibition System (Gray Reference Gray, Iversen, Iversen and Snyder1977) processes goal-conflict and amplifies behavioural inhibition/passive avoidance/defensive quiescence, attention, arousal, and negative bias (Gray & McNaughton Reference Gray and McNaughton2000; McNaughton & Corr Reference McNaughton and Corr2004). This is neurally distinct from the panicolytic-sensitive systems that mediate fight, flight, freezing, and withdrawal/active avoidance (Fig. 2), collectively known as the Fight-Flight-Freeze System (FFFS). Note that “fight” in this context is a defensive response and quite distinct as a behaviour from the predatory “attack” that is contrasted with “defense” in the target article – although in human personality questionnaire studies the relations between withdrawal, defensive attack, and predation are unclear (Corr Reference Corr2016). Contrary to the picture painted by De Dreu and Gross, it is anxiety rather than fear that is linked to the release of stress hormones like cortisol (see McNaughton Reference McNaughton1989, pp. 57–59); and, while 5HT_1A agonists are anxiolytic but not panicolytic, the serotonin system as a whole innervates and affects not only the Behavioural Inhibition System but also the withdrawal and the approach systems, with quite high-level consequences (Carver et al. Reference Carver, Johnson and Joormann2008).

Figure 2.

Hierarchical organization of approach, withdrawal, and behavioural inhibition (BIS) in terms of behaviour and neural level. Lower levels process small defensive distances; higher levels process greater ones (i.e., negative events that are more distant in space or time). Activation tends to spread through the whole system (double-headed black arrows), but strong activation of a higher level (e.g., avoidance) inhibits (single-headed arrows) the behavioural output from (but not the activation of) lower levels (e.g., escape). *Static postures that achieve withdrawal, conflict resolution, or approach, respectively. a = anterior; ad = anterodorsal; bl = basolateral; c = central; dl = dorsolateral; dm = dorsomedial; l = lateral; m = medial; OFC = orbital frontal cortex; p = posterior; PAG = periaqueductal grey; rm = rostromedial; vl = ventrolateral; vm = ventromedial. Reprinted, with permission, from McNaughton and Corr (Reference McNaughton and Corr2018).

For the same reasons, we think their picture of prefrontal control networks should be split and extended to subcortex. We agree that anxiety involves the inferior frontal gyrus (Shadli et al. Reference Shadli, McIntosh, Glue and McNaughton2015), basolateral amygdala, hippocampus, ventromedial orbital cortex (Fig. 2), and insula (Paulus & Stein Reference Paulus and Stein2006). However, we would add the posterior cingulate cortex, and, with risk assessment in particular (McNaughton & Corr Reference McNaughton and Corr2018), there is an important role for subcortical “survival circuits” (Mobbs & LeDoux Reference Mobbs and LeDoux2018) that include the dorso-lateral and ventro-lateral periaqueductal grey, anterior and lateral hypothalamus, and lateral septum (Motta et al. Reference Motta, Carobrez and Canteras2017). Critically, we see fear as neurally distinct, involving lateral orbital cortex, anterior cingulate, central amygdala, medial hypothalamus, and dorso-medial periaqueductal grey.

The three-system (approach, withdrawal, conflict) hierarchical neuropsychology we have described is also relevant to the trait considerations of section 3.6. “These systems mediate fluid moment-by-moment reactions to changing stimuli, with relatively stable person-specific sensitivities to these stimuli manifested in personality traits” (Corr et al. Reference Corr, DeYoung and McNaughton2013, p. 158) and are the basis for the Reinforcement Sensitivity Theory of personality (see Corr Reference Corr2008). Our perspective (avoiding the ambiguity of “reward”) suggests that attack (as a predatory approach tendency) and defense (functioning to allow withdrawal) likely depend on fundamentally similar hierarchical system architectures. Apparent prefrontal versus subcortical control differences between them likely depend on the usual difference in “motivational distance” in their eliciting situations. Initially, at least, a predator will be at a large appetitive distance from the prey, requiring extensive planning of its attack. Especially where an ambush is involved, the defensive response by the prey will be immediate and even undirected (at zero defensive distance). Impulse control also involves a balance between approach motivation and inhibition. The strength of inhibition can be affected by variations in conflict sensitivity (Gray & McNaughton Reference Gray and McNaughton2000) and in loss aversion (Tversky & Kahneman Reference Tversky and Kahneman1991); and approach can vary with the strength of delay discounting (Frost & McNaughton Reference Frost and McNaughton2017). The effects of traits on attack and defense clearly require a highly nuanced approach.