The dichotomy between following tradition versus innovating and the ability to switch between the two as described by Jagiello et al. is very much central to the pursuit and progress of science. Science is not discovery and invention alone. A substantial part of the practice of science consists of following tradition more or less unquestioned. Innovation in the right place and right proportion would facilitate healthy progress of science. Very parallel dichotomies exist at different levels of science. Thomas Kuhn's “normal science” versus “revolutionary science” (Kuhn, Reference Kuhn2020) is a dichotomy very parallel to bifocal stance theory (BST) albeit at a different level.

Jagiello et al. further recognize the importance of optimization between the two strategies. However, they have not sufficiently explored the underlying evolutionary principles of optimization. A working scientist needs to optimize tradition–innovation. The complex process of optimization is influenced by multiple factors including innate human behavior, culture, belief systems, social structure, economics, education, and prevalent academic systems. Much is being talked about flaws, biases, misconduct, reproducibility crisis, and conformity in today's science out of which conformity bias is directly related to BST (Padalia, Reference Padalia2014; Weatherall & O'Connor, Reference Weatherall and O'Connor2021) the roots of which are claimed to lie in evolved neuro-behavioral mechanisms (Germar, Albrecht, Voss, & Mojzisch, Reference Germar, Albrecht, Voss and Mojzisch2016; Morgan, Laland, Biele, Yoon, & Burke, Reference Morgan, Laland, Biele, Yoon and Burke2012; Watve, Reference Watve2017, Reference Watve2019). It is possible that insights into the details of the subconscious optimization models of researchers will allow us to design the education, training, institutionalization, and organization of science practice toward healthy, unbiased, and rapid progress of science.

An oversimplified model of optimizing tradition–invention is the snowdrift game (Sasaki & Okada, Reference Sasaki and Okada2015). If two cars are stuck because of a snowdrift, either or both the drivers can get down and shovel the snow to clear the path. If one is shoveling the snow, the other can very well sit in the car and enjoy the benefit, saving personal cost. Tradition is more likely to be a strategy of the driver sitting in the car. Innovator is the other driver taking the risk and efforts. Although innovation is considered desirable in science, the history of science has many examples where innovation faced rejection (Tröhler, Reference Tröhler2005; Weatherall & O'Connor, Reference Weatherall and O'Connor2021). Exploration or invention can be assumed to have higher risk, with some chance of giving high return. However, if successful, others can imitate the new trait and get the fruits of the invention. By the snowdrift dynamics, the cost–benefits of the alternative strategies are frequency dependent and the equilibrium lies in co-existence of both strategies. The reaction norm of frequency dependence will be the main determinant of the optimum combination. Real-life contexts are almost always more complex than classical theoretical games, but it is possible to refine the models contextually. The design of and the norms followed by academia would influence the parameters of frequency–cost–benefit relationship in a complex way and insightful research is needed to understand these reaction norms.

The other dimension of real-life complexity is that novelty or departure from tradition is not a binary but a continuous variable. Therefore, the question of interest is how much of innovation is optimum in a given context. Making innovation as a continuous variable demands an entirely different class of optimization models. There is likely to be an optimum novelty that maximizes the probability of acceptance by the community. So far there has been little effort to explore this possibility theoretically or empirically.

The third dimension of the complexity is differential individual propensity to innovate. Individual differences can arise partly from frequency-dependent selection. Not only humans have differences in individual creativity with a neurological basis (Beaty et al., Reference Beaty, Kenett, Christensen, Rosenberg, Benedek, Chen and Silvia2018), differential propensities as well as contextual flexibility are also known in animals. In animal models, the individual-learning strategy is often determined by their behavioral type (“personality”), age, social rank as well as social interactions (Coussi-Korbel & Fragaszy, Reference Coussi-Korbel and Fragaszy1995; Laland, Reference Laland2004). For example, in European starlings (Sturnus vulgaris), dominant individuals are less neophobic and thus learn a novel task quicker than subordinate group members (Boogert, Reader, & Laland, Reference Boogert, Reader and Laland2006) while in jackdaws (Corvus monedula), dominant individuals are far more likely to monopolize resources leaving little opportunity for subordinates to learn individually (Federspiel, Boeckle, von Bayern, & Emery, Reference Federspiel, Boeckle, von Bayern and Emery2019). However, in archerfish (Toxotes chatareus), only the personalities of individuals predict their learning propensity and not the social ranks (Jones, Spence-Jones, Webster, & Rendell, Reference Jones, Spence-Jones, Webster and Rendell2021). Within an individual, the choice between asocial learning, social learning, and innovation is flexible. In a given context, the choice depends on the cost of asocial learning, reliability, and usefulness of social information and the certainty of one's own information (Kendal, Coolen, van Bergen, & Laland, Reference Kendal, Coolen, van Bergen and Laland2005; Laland, Reference Laland2004). Thus, when learning by experience is costly (e.g., predator recognition and evasion, poisonous foods) or when the prior information possessed by the individual is unreliable (e.g., foraging in a heterogeneous or new patch), they resort to copying others. On the other hand, when both social and individual information is unusable, innovations are expected to occur. The same principles can apply to rituals where the perceived cost of learning through exploration or innovating changes with context, individual traits, and social standing. Certain trends in copying others also depend on the social ranking. In chimpanzees an innovation by high-ranking individuals is copied more readily by others as compared to one by a low-ranking individual (Biro et al., Reference Biro, Inoue-Nakamura, Tonooka, Yamakoshi, Sousa and Matsuzawa2003; Kendal et al., Reference Kendal, Hopper, Whiten, Brosnan, Lambeth, Schapiro and Hoppitt2015).

It is very likely that the ability to judge the context-specific parameters has evolved to very fine levels in humans which contributes further to the variance in innovativeness. But this complexity remains underexplored. The three dimensions of complexity, namely frequency dependence, continuous nature of innovation, and individual propensity need to be investigated using theoretical as well as empirical tools. The field of science and the process of research has not been studied from a behavioral perspective barring a few attempts (Chapman et al., Reference Chapman, Bicca-Marques, Calvignac-Spencer, Fan, Fashing, Gogarten and Stenseth2019; Watve, Reference Watve2017, Reference Watve2019). BST is likely to be the platform on which to initiate the process. Understanding this complexity, at least in part, will help us design better science education, research training, research environment, and science publishing.