The motivation literature typically distinguishes between process theories and content theories. Process theories seek to provide a mechanistic explanation of behavior, specifying a mechanism between input (stimulus) and output (behavior) that includes motivation constructs such as goals. Content theories address questions about what our fundamental goals are (e.g., autonomy, competence, and connectedness in self-determination theory; Ryan & Deci, Reference Ryan and Deci2017), or they apply existing process theories to a specific type of goal (e.g., achievement, affiliation, power; Murray, Reference Murray1938). Unlike content theories, existing process theories have already taken steps toward opening the black box that sits between observable stimulus input and behavioral output. The critique of M&J therefore seems to be more directed toward content theories than toward process theories. Alternatively, their critique could be directed toward process theories, suggesting that these theories do not unpack their mechanisms in sufficient detail, that is, at a low enough level of analysis.
The authors go on to propose a way to do this unpacking. Despite cautioning that their exercise is not a reductionist attempt, they appear to promote three types of reductions. I will argue how each of these types poses risks for throwing away the baby with the bathwater.
A first type of reduction is a shift from high-level (abstract) motivational constructs to low-level (concrete) motivational constructs. For instance, they propose replacing the goal for “competence” with the goal for “information seeking.” The ideal seems to shift to motivational constructs that are as close as possible to the behavior itself. Thus, they argue that information-seeking behavior is caused by the goal to have information rather than by the goal for competence. They further propose that such a low-level goal is caused by a gap in this goal. Thus, the goal to have information is caused by uncertainty, that is, a gap between a current and desired level of information.
The ultimate step in this shift toward low-level constructs is the shift from more extrinsic to more intrinsic motivation. People seek information because they value the act of information seeking (or at least the immediate outcome of this act: information) rather than because this information is instrumental for reaching other goals. Such a shift seems to downplay the role of extrinsic motivation, however. There may certainly be cases in which people act for the sake of it, but much of our behavior, including information seeking, is done to reach other goals. The explanatory power of existing process theories resides in the fact that they propose a goal hierarchy in which behavioral goals that are the proximal causes of behavior can be considered as subordinate goals that have a high expectancy for reaching other, superordinate goals.
This feature is preserved in several existing process theories, including in my own goal-directed theory of behavior causation (Moors, Reference Moors2022; Moors, Boddez, & De Houwer, Reference Moors, Boddez and De Houwer2017). Here, a gap between a high-level goal and a current state activates the goal to reduce this gap. This can be done by different types of behaviors (i.e., assimilation), but also by adjusting the high-level goal (i.e., accommodation) or by changing the interpretation of the current state (i.e., immunization). To illustrate, the gap between the current state and the high-level goal to become popular activates the goal to reduce this gap. One way to reduce this gap is via behavior (e.g., wearing nice cloths, making jokes; i.e., assimilation), another is to give up the goal to become popular (i.e., accommodation), and still another is to reinterpret the current state as one in which you are already popular (i.e., immunization).
It might be argued that the high-level goals proposed in content theories such as the goals for autonomy and competence are not well chosen, but then the solution would be to come up with better ones, not to do away with high-level goals altogether. An alternative solution would be to view the goals for autonomy and competence as meta-goals that are at the service of, or assist in, the attainment of other, low-level goals. In this vein, the goal for autonomy can be considered as the goal to be allowed to choose one's own (low-level) goals and the goal for competence or control can be considered as the goal to achieve these (low-level) goals.
A second type of reduction that M&J seem to promote consists in a shift from motivation to cognition. The authors admit that in their computational process model, there are still rewards, which are representations of valued outcomes and hence motivational constructs. However, once the unpacking of the black box has arrived at its most concrete level of motivation, the authors argue that it makes little sense to continue calling this motivation. This reveals that the ideal to which they aspire is to ultimately reduce motivation to cognition. This is reminiscent of the attempt of predictive processing theory to reduce the explanation of behavior (and other phenomena such as perception and affect) to the confirmation and disconfirmation of expectations (Clark, Reference Clark2013; Friston, Reference Friston2009).
A third type of reduction that the authors seem to promote consists in gradually explaining away content or semantics. In standard mechanistic explanations in psychology, mechanisms between stimulus input and behavioral output are composed of representations with a content and a format (i.e., the structural parts) and operations acting on these representations (i.e., the activities or working parts) (Bechtel, Reference Bechtel2008). The authors’ ideal seems to be to reduce mechanistic explanations that consist of both contentful representations and operations to explanations that consist only of operations.
In conclusion, even if existing process theories of motivation have already made progress in unpacking the black box, it may be argued that there is always room for further unpacking at lower levels of analysis. Whether this shift to lower levels of analysis should include the rejection of high-level goals, the reduction of motivation to cognition, and the evolution toward content-less mechanistic explanations is open to debate, as is the level of analysis that will prove to be most fruitful for predicting and regulating behavior (Karoly, Reference Karoly1999).
You have
Access
The motivation literature typically distinguishes between process theories and content theories. Process theories seek to provide a mechanistic explanation of behavior, specifying a mechanism between input (stimulus) and output (behavior) that includes motivation constructs such as goals. Content theories address questions about what our fundamental goals are (e.g., autonomy, competence, and connectedness in self-determination theory; Ryan & Deci, Reference Ryan and Deci2017), or they apply existing process theories to a specific type of goal (e.g., achievement, affiliation, power; Murray, Reference Murray1938). Unlike content theories, existing process theories have already taken steps toward opening the black box that sits between observable stimulus input and behavioral output. The critique of M&J therefore seems to be more directed toward content theories than toward process theories. Alternatively, their critique could be directed toward process theories, suggesting that these theories do not unpack their mechanisms in sufficient detail, that is, at a low enough level of analysis.
The authors go on to propose a way to do this unpacking. Despite cautioning that their exercise is not a reductionist attempt, they appear to promote three types of reductions. I will argue how each of these types poses risks for throwing away the baby with the bathwater.
A first type of reduction is a shift from high-level (abstract) motivational constructs to low-level (concrete) motivational constructs. For instance, they propose replacing the goal for “competence” with the goal for “information seeking.” The ideal seems to shift to motivational constructs that are as close as possible to the behavior itself. Thus, they argue that information-seeking behavior is caused by the goal to have information rather than by the goal for competence. They further propose that such a low-level goal is caused by a gap in this goal. Thus, the goal to have information is caused by uncertainty, that is, a gap between a current and desired level of information.
The ultimate step in this shift toward low-level constructs is the shift from more extrinsic to more intrinsic motivation. People seek information because they value the act of information seeking (or at least the immediate outcome of this act: information) rather than because this information is instrumental for reaching other goals. Such a shift seems to downplay the role of extrinsic motivation, however. There may certainly be cases in which people act for the sake of it, but much of our behavior, including information seeking, is done to reach other goals. The explanatory power of existing process theories resides in the fact that they propose a goal hierarchy in which behavioral goals that are the proximal causes of behavior can be considered as subordinate goals that have a high expectancy for reaching other, superordinate goals.
This feature is preserved in several existing process theories, including in my own goal-directed theory of behavior causation (Moors, Reference Moors2022; Moors, Boddez, & De Houwer, Reference Moors, Boddez and De Houwer2017). Here, a gap between a high-level goal and a current state activates the goal to reduce this gap. This can be done by different types of behaviors (i.e., assimilation), but also by adjusting the high-level goal (i.e., accommodation) or by changing the interpretation of the current state (i.e., immunization). To illustrate, the gap between the current state and the high-level goal to become popular activates the goal to reduce this gap. One way to reduce this gap is via behavior (e.g., wearing nice cloths, making jokes; i.e., assimilation), another is to give up the goal to become popular (i.e., accommodation), and still another is to reinterpret the current state as one in which you are already popular (i.e., immunization).
It might be argued that the high-level goals proposed in content theories such as the goals for autonomy and competence are not well chosen, but then the solution would be to come up with better ones, not to do away with high-level goals altogether. An alternative solution would be to view the goals for autonomy and competence as meta-goals that are at the service of, or assist in, the attainment of other, low-level goals. In this vein, the goal for autonomy can be considered as the goal to be allowed to choose one's own (low-level) goals and the goal for competence or control can be considered as the goal to achieve these (low-level) goals.
A second type of reduction that M&J seem to promote consists in a shift from motivation to cognition. The authors admit that in their computational process model, there are still rewards, which are representations of valued outcomes and hence motivational constructs. However, once the unpacking of the black box has arrived at its most concrete level of motivation, the authors argue that it makes little sense to continue calling this motivation. This reveals that the ideal to which they aspire is to ultimately reduce motivation to cognition. This is reminiscent of the attempt of predictive processing theory to reduce the explanation of behavior (and other phenomena such as perception and affect) to the confirmation and disconfirmation of expectations (Clark, Reference Clark2013; Friston, Reference Friston2009).
A third type of reduction that the authors seem to promote consists in gradually explaining away content or semantics. In standard mechanistic explanations in psychology, mechanisms between stimulus input and behavioral output are composed of representations with a content and a format (i.e., the structural parts) and operations acting on these representations (i.e., the activities or working parts) (Bechtel, Reference Bechtel2008). The authors’ ideal seems to be to reduce mechanistic explanations that consist of both contentful representations and operations to explanations that consist only of operations.
In conclusion, even if existing process theories of motivation have already made progress in unpacking the black box, it may be argued that there is always room for further unpacking at lower levels of analysis. Whether this shift to lower levels of analysis should include the rejection of high-level goals, the reduction of motivation to cognition, and the evolution toward content-less mechanistic explanations is open to debate, as is the level of analysis that will prove to be most fruitful for predicting and regulating behavior (Karoly, Reference Karoly1999).
Financial support
Preparation of this article was supported by grant C14/23/062 of the Research Fund of KU Leuven.
Competing interests
None.