Motivation theory has long been a battleground, as evidenced in part by its long history of competing lists of basic drives and motives (or wants and needs), with no strong method for evaluating such lists. One perennially attempted solution has been to reduce motivation to cognition. Murayama and Jach (henceforth M&J) provide one of the more intelligent and reasonable efforts of this sort. Perhaps people do not have wants or needs at all – instead they have computational processes.
We are pleased to see that after discussing how to get rid of concepts of motivation, M&J conclude that they are not eliminating motivation after all. They hope to add a new level of analysis that can improve our understanding. This is promising. Computational processes might not replace motivation after all but elaborate how motivations work.
At points they do say that motivation processes can be re-framed as computational processes, such as by seeking rewarding information. But, crucially – what makes a reward rewarding? The answer is that the person wants or needs it. (This applies to information also.) Motivation is something inside the person (or other animal) that makes the reward appealing. Identical stimuli can be highly rewarding to some people but completely indifferent to others. The rewardingness is not in the stimulus but in the person, or at best in its relationship to the person. Computers can perform calculations faster and better than humans, but information is not rewarding to the computer – precisely because the computer lacks motivation. By and large, cognition serves motivation, but both are intertwined and cannot replace each other. Motivation without cognition would be endless frustration. Cognition without motivation would not know what to do.
We think motivation is fundamental to psychology, because it is most closely linked to sustaining life. Organisms evolved to want food, safety, sleep, sex, and the like, and these motivations helped them to survive and reproduce. Computers, contrast, are indifferent to whether they survive or reproduce. Cognition can serve motivation by helping the animal understand how to obtain the resources it needs to sustain life. Human cognition can even inhibit behaviors leading to immediate but problematic rewards in order to obtain delayed high-quality rewards.
Much of M&J's analysis relies on information seeking. This seems an atypical example that is exceptionally conducive to attempts to reduce motivation to cognition. Desires for sex, power, social status, money, or even fame would seemingly pose a more formidable challenge. It is not clear to us how to frame such desires as computational processes. Is the desire to copulate at bottom a search for rewarding information? And if so, what makes copulation rewarding? Sexual intercourse is not the filling of a couple's knowledge gap, except perhaps for their first time. Saving money for the proverbial rainy day is likewise not about filling gaps in knowledge.
Moreover, the existence of an information gap is not enough to produce motivation. There are many things that we do not know, and we know that we do not know them, but we are not motivated to learn them, such as the phone numbers of far-off strangers, or the middle names of thousands of deceased people. To their credit, M&J acknowledge that the reward-learning framework fails to state what kind of information is rewarding (or why). This is a crucial point that is not a detail but indicates a central shortcoming of the entire approach.
Ultimately, something crucial is missing. The desire for information is not the most basic human drive, from which all other seemingly motivated patterns can be deduced. In our view, motivation evolved as a subjective craving for things that contributed to biological success at survival and reproduction. Yes, having good information helps those things. But more fundamentally, sex, social status, power, food, safety, and similar things are keys to survival and reproduction. Some of these might not be included under their umbrella concept of “higher-order” motivations. But we wish they would define what differentiates higher-order motivations from others and perhaps provide a short list.
Child development research may also give pause to those who seek to explain motivation with cognition. Redding, Morgan, and Harmon (Reference Redding, Morgan and Harmon1988) found that task persistence measures correlated more weakly with cognitive measures among older than younger children, and they concluded that motivation and cognition may become increasingly separate as children grow older. Likewise, babies show exploratory behavior and selective interest – so motivation operates long before there is much in the way of prefrontal computational process.
Furthermore, the hallmarks of depression (i.e., lack of interest, slowness in decision-making, difficulty paying attention, poor concentration, and passive inactivity) seem to be explained by the interaction between motivational and cognitive processes (Grahek, Shenhav, Musslick, Krebs, & Koster, Reference Grahek, Shenhav, Musslick, Krebs and Koster2019). Put another way, the relationship between reward processing and cognitive control across different clinical population suggests that these two systems remain independent even if they are strongly intertwined. These current perspectives are opposed to that proposed by M&J.
M&J are correct to note the challenges and frustrating problems in motivation theory. Elucidating computational processes may be valuable for filling gaps. But trying to replace motivation theory with cognitive theory has never worked.
You have
Access
Motivation theory has long been a battleground, as evidenced in part by its long history of competing lists of basic drives and motives (or wants and needs), with no strong method for evaluating such lists. One perennially attempted solution has been to reduce motivation to cognition. Murayama and Jach (henceforth M&J) provide one of the more intelligent and reasonable efforts of this sort. Perhaps people do not have wants or needs at all – instead they have computational processes.
We are pleased to see that after discussing how to get rid of concepts of motivation, M&J conclude that they are not eliminating motivation after all. They hope to add a new level of analysis that can improve our understanding. This is promising. Computational processes might not replace motivation after all but elaborate how motivations work.
At points they do say that motivation processes can be re-framed as computational processes, such as by seeking rewarding information. But, crucially – what makes a reward rewarding? The answer is that the person wants or needs it. (This applies to information also.) Motivation is something inside the person (or other animal) that makes the reward appealing. Identical stimuli can be highly rewarding to some people but completely indifferent to others. The rewardingness is not in the stimulus but in the person, or at best in its relationship to the person. Computers can perform calculations faster and better than humans, but information is not rewarding to the computer – precisely because the computer lacks motivation. By and large, cognition serves motivation, but both are intertwined and cannot replace each other. Motivation without cognition would be endless frustration. Cognition without motivation would not know what to do.
We think motivation is fundamental to psychology, because it is most closely linked to sustaining life. Organisms evolved to want food, safety, sleep, sex, and the like, and these motivations helped them to survive and reproduce. Computers, contrast, are indifferent to whether they survive or reproduce. Cognition can serve motivation by helping the animal understand how to obtain the resources it needs to sustain life. Human cognition can even inhibit behaviors leading to immediate but problematic rewards in order to obtain delayed high-quality rewards.
Much of M&J's analysis relies on information seeking. This seems an atypical example that is exceptionally conducive to attempts to reduce motivation to cognition. Desires for sex, power, social status, money, or even fame would seemingly pose a more formidable challenge. It is not clear to us how to frame such desires as computational processes. Is the desire to copulate at bottom a search for rewarding information? And if so, what makes copulation rewarding? Sexual intercourse is not the filling of a couple's knowledge gap, except perhaps for their first time. Saving money for the proverbial rainy day is likewise not about filling gaps in knowledge.
Moreover, the existence of an information gap is not enough to produce motivation. There are many things that we do not know, and we know that we do not know them, but we are not motivated to learn them, such as the phone numbers of far-off strangers, or the middle names of thousands of deceased people. To their credit, M&J acknowledge that the reward-learning framework fails to state what kind of information is rewarding (or why). This is a crucial point that is not a detail but indicates a central shortcoming of the entire approach.
Ultimately, something crucial is missing. The desire for information is not the most basic human drive, from which all other seemingly motivated patterns can be deduced. In our view, motivation evolved as a subjective craving for things that contributed to biological success at survival and reproduction. Yes, having good information helps those things. But more fundamentally, sex, social status, power, food, safety, and similar things are keys to survival and reproduction. Some of these might not be included under their umbrella concept of “higher-order” motivations. But we wish they would define what differentiates higher-order motivations from others and perhaps provide a short list.
Child development research may also give pause to those who seek to explain motivation with cognition. Redding, Morgan, and Harmon (Reference Redding, Morgan and Harmon1988) found that task persistence measures correlated more weakly with cognitive measures among older than younger children, and they concluded that motivation and cognition may become increasingly separate as children grow older. Likewise, babies show exploratory behavior and selective interest – so motivation operates long before there is much in the way of prefrontal computational process.
Furthermore, the hallmarks of depression (i.e., lack of interest, slowness in decision-making, difficulty paying attention, poor concentration, and passive inactivity) seem to be explained by the interaction between motivational and cognitive processes (Grahek, Shenhav, Musslick, Krebs, & Koster, Reference Grahek, Shenhav, Musslick, Krebs and Koster2019). Put another way, the relationship between reward processing and cognitive control across different clinical population suggests that these two systems remain independent even if they are strongly intertwined. These current perspectives are opposed to that proposed by M&J.
M&J are correct to note the challenges and frustrating problems in motivation theory. Elucidating computational processes may be valuable for filling gaps. But trying to replace motivation theory with cognitive theory has never worked.
Financial support
Not applicable.
Competing interest
N. A. and R. F. B. declare that they have no conflict of interest.