Lieder and Griffiths argue that cognitive modeling should proceed from the assumption that mental processes are optimizing, but in a cognitively efficient manner: “cognitive mechanisms [should be expected to] trade-off accuracy against effort in an adaptive, nearly optimal manner.” Although there is much that speaks in favor of this kind of resource-rational analysis, the authors underemphasize the evolutionary foundations of resource-rational analysis. This is problematic, as emphasizing these evolutionary foundations allows resource-rational analysis to be expanded and strengthened. Specifically, by taking into consideration the fact that efficiency considerations are a key driver of the evolution of human and animal minds (Schulz Reference Schulz2018), it becomes clearer: (1) why it is reasonable to assume that cognitive mechanisms trade off accuracy against effort, (2) how this trade-off occurs, and (3) how to overcome some of the challenges of this framework. Consider these three points in turn.
First, without placing the appeal to evolutionary biology front and center, it is not clear why resource-rational analysis works. Why should it be presumed that humans evolved in such a way as to “trade off accuracy against effort in an adaptive, nearly optimal manner”? In particular, why didn't humans evolve so as to “satisfice” without ever focusing on what's optimal (as has been suggested e.g., by Gigerenzer & Selten Reference Gigerenzer and Selten2001)? To answer these questions, it is necessary to consider the biological costs and benefits of (a) accurate decision-making, and (b) time- and cognitive resource-hungry-decision making. That is, it needs to be specified why accurate decision-making is biologically advantageous, and why these biological advantages are tempered by the cognitive and temporal costs that come from accurate decisions. Why, exactly, is an organism's fitness increased by relying on accurate cognitive processes, and decreased by relying on ones that take much time, concentration, and attention? It is just not obvious why an organism's expected reproductive success is affected by the accuracy and efficiency of cognitive processes. A step in the direction of an answer lies in the fact that it is reasonable to see cognitive efficiency and neurobiological efficiency as correlated, so that resource-rational decision-making allows organisms to save energy in maintaining and updating their central nervous system – a fact that is evolutionarily important (see Schulz Reference Schulz2018, for more on this). In general, it is only by providing a detailed account of the biological costs and benefits of resource-rational decision-making that Lieder and Griffiths's framework can be put on a plausible foundation and be properly distinguished from rival frameworks like that of Gigerenzer et al.
Second, a closer look at the evolutionary pressures on cognitive processing allows for improved predictions about when the tradeoff between accuracy and efficiency is resolved in which way. A good illustration of this is the authors’ point that “people often think only about which subgoal to pursue next and how to achieve it […]. This is suboptimal from the perspective of expected utility theory […]. The resource-rationality framework can reconcile this tension by pointing out that goal-directed planning affords many computational simplifications that make good decision-making tractable.” This, however, leaves it open when, exactly, people should be expected to rely on subgoals in their decision-making – and when they should be expected to rely on a more general goal of maximizing their expected utilities (or some such). An appeal to considerations from evolutionary biology can help answer these questions. For example, for many mammals, helping offspring in need is a choice that often needs to be made quickly and which responds to a unique set of environmental variables (viz., that the organism in need is one's offspring). From an evolutionary biological perspective, it can thus be predicted that this choice will be driven by a separate subgoal for helping offspring in need (see also Piccinini & Schulz Reference Piccinini and Schulz2019). By contrast, the choices humans need to make in order to navigate through their complex social environments – such as who to marry or what kinds of coalitions to join – respond to a wide range of variables (who is related to who, who is in a collation with who, who has which kinds of social statuses, etc.), but can typically be made less efficiently (in a matter of days rather than seconds). From an evolutionary biological perspective, it can thus be predicted that these choices will not be driven by separate subgoals, but by a more general goal of doing the best one can in society (or some such) (see also Schulz Reference Schulz2018, Ch. 8). In this way, an evolutionary perspective is useful for making predictions about the details of resource-rational analysis that can then be tested further.
Third, an evolutionary perspective allows resource-rational analysis to overcome some of the challenges it faces. Lieder and Griffiths note that “people's performance during the process of adaptation to a new environment” is difficult to analyze with their framework. However, from an evolutionary perspective, an organism's ability to efficiently adapt to new environments is just as much under selection as their ability to make decisions in a given environment. For example, there are good evolutionary biological reasons for thinking that more computationally intensive cognitive processes have an easier time adjusting to new environments, but that they pay a price in terms of the time, concentration, attention, etc., they need for their execution (Schulz Reference Schulz2018). By taking an evolutionary perspective more explicitly into account, it would thus become possible to broaden resource-rational analysis so that it can be applied to questions about “people's performance during the process of adaptation to a new environment” after all.
For these reasons, a fully compelling resource-rational analysis would profit greatly from being integrated more tightly with a more explicit evolutionary psychological foundation. This should thus be the next step in the development of this otherwise compelling theoretical framework.
Lieder and Griffiths argue that cognitive modeling should proceed from the assumption that mental processes are optimizing, but in a cognitively efficient manner: “cognitive mechanisms [should be expected to] trade-off accuracy against effort in an adaptive, nearly optimal manner.” Although there is much that speaks in favor of this kind of resource-rational analysis, the authors underemphasize the evolutionary foundations of resource-rational analysis. This is problematic, as emphasizing these evolutionary foundations allows resource-rational analysis to be expanded and strengthened. Specifically, by taking into consideration the fact that efficiency considerations are a key driver of the evolution of human and animal minds (Schulz Reference Schulz2018), it becomes clearer: (1) why it is reasonable to assume that cognitive mechanisms trade off accuracy against effort, (2) how this trade-off occurs, and (3) how to overcome some of the challenges of this framework. Consider these three points in turn.
First, without placing the appeal to evolutionary biology front and center, it is not clear why resource-rational analysis works. Why should it be presumed that humans evolved in such a way as to “trade off accuracy against effort in an adaptive, nearly optimal manner”? In particular, why didn't humans evolve so as to “satisfice” without ever focusing on what's optimal (as has been suggested e.g., by Gigerenzer & Selten Reference Gigerenzer and Selten2001)? To answer these questions, it is necessary to consider the biological costs and benefits of (a) accurate decision-making, and (b) time- and cognitive resource-hungry-decision making. That is, it needs to be specified why accurate decision-making is biologically advantageous, and why these biological advantages are tempered by the cognitive and temporal costs that come from accurate decisions. Why, exactly, is an organism's fitness increased by relying on accurate cognitive processes, and decreased by relying on ones that take much time, concentration, and attention? It is just not obvious why an organism's expected reproductive success is affected by the accuracy and efficiency of cognitive processes. A step in the direction of an answer lies in the fact that it is reasonable to see cognitive efficiency and neurobiological efficiency as correlated, so that resource-rational decision-making allows organisms to save energy in maintaining and updating their central nervous system – a fact that is evolutionarily important (see Schulz Reference Schulz2018, for more on this). In general, it is only by providing a detailed account of the biological costs and benefits of resource-rational decision-making that Lieder and Griffiths's framework can be put on a plausible foundation and be properly distinguished from rival frameworks like that of Gigerenzer et al.
Second, a closer look at the evolutionary pressures on cognitive processing allows for improved predictions about when the tradeoff between accuracy and efficiency is resolved in which way. A good illustration of this is the authors’ point that “people often think only about which subgoal to pursue next and how to achieve it […]. This is suboptimal from the perspective of expected utility theory […]. The resource-rationality framework can reconcile this tension by pointing out that goal-directed planning affords many computational simplifications that make good decision-making tractable.” This, however, leaves it open when, exactly, people should be expected to rely on subgoals in their decision-making – and when they should be expected to rely on a more general goal of maximizing their expected utilities (or some such). An appeal to considerations from evolutionary biology can help answer these questions. For example, for many mammals, helping offspring in need is a choice that often needs to be made quickly and which responds to a unique set of environmental variables (viz., that the organism in need is one's offspring). From an evolutionary biological perspective, it can thus be predicted that this choice will be driven by a separate subgoal for helping offspring in need (see also Piccinini & Schulz Reference Piccinini and Schulz2019). By contrast, the choices humans need to make in order to navigate through their complex social environments – such as who to marry or what kinds of coalitions to join – respond to a wide range of variables (who is related to who, who is in a collation with who, who has which kinds of social statuses, etc.), but can typically be made less efficiently (in a matter of days rather than seconds). From an evolutionary biological perspective, it can thus be predicted that these choices will not be driven by separate subgoals, but by a more general goal of doing the best one can in society (or some such) (see also Schulz Reference Schulz2018, Ch. 8). In this way, an evolutionary perspective is useful for making predictions about the details of resource-rational analysis that can then be tested further.
Third, an evolutionary perspective allows resource-rational analysis to overcome some of the challenges it faces. Lieder and Griffiths note that “people's performance during the process of adaptation to a new environment” is difficult to analyze with their framework. However, from an evolutionary perspective, an organism's ability to efficiently adapt to new environments is just as much under selection as their ability to make decisions in a given environment. For example, there are good evolutionary biological reasons for thinking that more computationally intensive cognitive processes have an easier time adjusting to new environments, but that they pay a price in terms of the time, concentration, attention, etc., they need for their execution (Schulz Reference Schulz2018). By taking an evolutionary perspective more explicitly into account, it would thus become possible to broaden resource-rational analysis so that it can be applied to questions about “people's performance during the process of adaptation to a new environment” after all.
For these reasons, a fully compelling resource-rational analysis would profit greatly from being integrated more tightly with a more explicit evolutionary psychological foundation. This should thus be the next step in the development of this otherwise compelling theoretical framework.