Main

Under some conditions, Wolf et al.1 find consistent behavioural differences between individuals that reproduce early in life and those that delay reproduction to explore their habitats instead to enhance future reproduction. Non-explorers that reproduce early in life later become bold and aggressive, whereas exploratory individuals with greater future reproductive potential are shy and unaggressive. These differences are caused by asset protection5 where individuals with greater future fitness take fewer risks that would jeopardize that fitness.

Asset protection, however, is a negative feedback process that, given time, makes individuals more alike, not less. In Clark’s original asset protection paper5, many decisions are made over an animal’s lifespan. Over time, individuals tend towards similar behaviour, despite any initial differences in assets, because those with assets take few risks and acquire little new fitness. Those without high assets take more risks and (unless they die trying) acquire new fitness assets that become worth protecting.

If, in the model of Wolf et al., individuals experience many hawk–dove encounters, successful hawks would eventually accumulate enough fitness for playing dove to become their optimal behaviour. Given time to accrue new assets, behavioural types would converge. Two particular conditions that could prevent this convergence are: animals with very short lives might not have time to change their assets sufficiently to cause changes in behaviour; and early life-history choices can have such large fitness consequences that subsequent bold and aggressive behaviour has relatively little influence on assets. Notably, these conditions do not seem to fit the maintenance of stable personalities in long-lived organisms such as humans.

The model of Wolf et al. requires bold/aggressive contexts not to dominate one another in fitness consequences, otherwise the negative feedback of asset protection will apply at this smaller scale (Supplementary Fig. 2 of ref. 1: in the square in which behavioural correlations could evolve, there is a wedge-shaped region without correlation between the hawk–dove and predator games). We reproduced their model and found that, in this region, thorough explorers are less aggressive than non-explorers, but no one is bold. Without the hawk–dove game, explorers would be shy and non-explorers bold, but when the hawk–dove game has sufficiently higher fitness consequences than the boldness game, all individuals are shy to eliminate the risk of dying before the fitness windfall from the hawk–dove game. This is the asset-protection principle, working on the scale of the low-fitness behavioural contexts, producing behavioural inconsistency, unless the contexts do not dominate one another.

An alternative way of explaining behavioural consistency and correlations is through positive (not negative) feedback. For example, if thorough explorers gain assets (energy, size, knowledge) that improve their abilities to escape predators or to win fights, then we might find positive correlations between exploration, boldness and aggressiveness. Additional behaviour would positively feed back on state, maintaining differences in assets and behavioural types. What is needed next is a unified modelling framework in which both negative and positive state feedback, as well as other mechanisms, can be compared.