Table 3. Example stories.

Concept	Initial narrative of optimality	Evidence of suboptimality (strong but unviable null model)	Restoration of optimality through the inclusion of an ad hoc tradeoff	Alternative benchmark narrative (strong and viable null model)
Criticality (Fontenele et al., 2019; Wilting and Priesemann, 2019; Nanda et al., 2023)	Brain activity always and exactly balances between order and disorder. This allows it to optimize information transmission and storage.	Brain activity does not always or exactly balance between order and disorder.	Brain activity optimizes the tradeoffs between the benefits of criticality and the competing benefits of flexibility or stability.	Brain activity avoids the extremes of overinhibition and overexcitation and is not optimal over and above this avoidance-of-extremes baseline.
Predictive coding (Sun and Firestone, 2020; Van de Cruys et al., 2020; Seth et al., 2020; Cao, 2020)	Brain activity aims to optimally predict incoming sensory input.	Brain activity optimally predicts sensory input in dark and quiet spaces. Despite this, animals tend not to seek out such spaces.	Brain activity aims to optimize the tradeoffs between predictions that are accurate and predictions that are motivational.	Brain activity reacts to sensory input but does not aim to optimally predict this input.
Wiring minimization (Markov et al., 2013; Bullmore and Sporns, 2012; Rubinov, 2016)	Brain-network structure globally minimizes wiring cost and therefore optimizes wiring economy.	Brain-network structure does not globally minimize wiring cost.	Brain-network structure optimizes the tradeoffs between wiring cost and communication efficiency.	Brain networks have long connections that enable specific sensory-motor function but do not optimize global communication.