Cognitive Redistribution Theory: Reconciling Apparently Conflicting Findings on Cognitive Offloading and Operationalizing Cognitive Sovereignty in the Second Brain

The rapid advancement of generative AI has brought human cognition to a new turning point. Cognitivesub-processes such as organizing information, comparing alternatives, and structuring arguments, which this papercollectively terms “cognitive chores,” can now be delegated to AI assistants. This paper proposes “Cognitive RedistributionTheory” to organize findings in the cognitive offloading literature that appear contradictory but are infact mutually consistent because they measure different cognitive outcomes. The “Google Effect” [Sparrow 11]showed that the expectation of external storage reduces memory for the externally stored information itself, whereas“Saving-Enhanced Memory” [Storm 15] showed that saving previously studied information enhances memory forsubsequently studied new information. The core proposition is that externalization triggers not memory “degradation”but a conditional “strategic redistribution” of cognitive resources, formulated as a two-stage mechanism:(1) reduction of encoding priority for externalized information, and (2) facilitation of subsequent cognitive activitythrough reduced proactive interference and the reallocation of freed resources. We identify three boundary conditionsfor beneficial redistribution—reliability of the externalization process, sufficient interference, and metacognitivecontrol—grounded in the saving experiments [Storm 15] and metacognition research [Grinschgl 21, Risko 16], withconverging evidence from the photo-taking impairment effect [Henkel 14] and extension studies [Fellers 24, Runge19]. The direct empirical basis concerns the external-memory role of such systems; the interactive, content-generatingrole of generative AI is treated as a boundary of scope and a future extension, while the metacognitive condition andthe sovereignty criteria remain applicable to it. Drawing on Klein and Klein’s “Cognitive Sovereignty” [Klein 25],we propose three operational criteria—degree of metacognitive engagement, replaceability (rollback capability), andlearning continuity—and provide design principles for AI-integrated Second Brain systems that extend cognitionwithout undermining cognitive sovereignty.