The claim
Psychological egoism has long been charged with equivocation. Read one way, that every motive is the agent’s own, it is trivially true. Read the other way, that every motive aims at the agent’s own good, it needs argument. The owner’s thesis (CLAIM-491) faces the same risk at a new joint, the extent of the self. If the self is whatever the agent identifies with, and identifying with something means caring about it, then “all action serves an extended self” says only “all action serves what the agent cares about”.
The thesis avoids this only by tying the self’s extent to a criterion that can diverge from caring, and predicting that motivation follows the criterion. Candidates:
- perceived self–other overlap (Aron’s scale);
- identity fusion;
- salient self-categorization;
- relatedness.
What it does not say
- Not that the thesis is false. It says what the thesis must add to be testable.
- Not that every criterion of the self is circular. Self-categorization theory’s salience by fit, for instance, is fixed by the comparative context, not by what the agent values (THEORY-1114).
- Not about caring in a reflective sense. “Cares about” here is what the agent is moved to promote. Read as reflective endorsement, a self defined by caring would leave out compulsive and akratic motives, and the thesis would then be false of those motives rather than empty.
Note of 2026-10-29: candidate criteria, tested
Works read in place of unavailable papers test the candidate criteria this objection lists.
- Self-definition and self-investment come apart (Leach et al., NOTE-1262). Identification splits into self-definition (seeing oneself as like the average member) and self-investment (satisfaction, solidarity, centrality). They correlate .61 to .77, but faced with their group’s misdeeds, self-definition predicted guilt and investment predicted defending the group. Self-stereotyping is a criterion of the self’s extent that separates from caring.
- The overlap circles are weaker than supposed. In Leach et al. the self–group circles correlated more with satisfaction and solidarity than with self-stereotyping, against the authors’ prediction. Branand, Mashek and Aron (NOTE-1276) report that the IOS scale is validated as subjective closeness. Their cognitive signs of inclusion are better candidates, because caring alone does not predict them: slower self-judgements where a partner differs, and self–partner confusions in memory.
- Implicitly, partly separable (Pinter and Greenwald, NOTE-1319). An IAT of self–group association and one of liking correlated .48 to .66, and their explicit versions .71.
- Salience moves without caring (Smith and Mackie, NOTE-1285; Spears and Postmes, NOTE-1280). Emotions and influence shift with whichever category is made salient by instruction or context. That fixes the self’s level independently of what one values. But the highly identified reappraise their group’s misdeeds to avoid guilt, so caring feeds back.
- Inside social identity theory (Scheepers and Ellemers, NOTE-1315). Tajfel defined social identity to include “the value or emotional significance attached to that membership”, so the theory’s social self already contains caring, though its measures separate the two. The chapter reports, without a source, that where they are separated, commitment predicts behaviour better. If so, motivation follows caring rather than the self’s extent.
Note of 2026-10-29: the criterion in time
- Future-self overlap predicted where caring did not (Hershfield, NOTE-1259). Similarity to one’s self in ten years, on overlapping circles, predicted patient choice, and directly reported liking and caring did not. But the caring reports were near ceiling, so the separation is suggested, not shown.
- The objection reappears in the theory of welfare (Luper, SEP, NOTE-1294). If posthumous harm is the thwarting of desires, and any satisfied desire counts as one’s good, legacy-seeking is self-interest by definition. The same triviality, one level down.