Cognitive offloading: when tools do more of the thinking.

The research line begins well before modern AI. The foundational review by Risko and Gilbert (2016) traced cognitive offloading across decades of work on calendars, notes, gestures, calculators, and external memory aids, showing that people consistently trade internal effort for tool use when the tool is reliable and convenient. The pattern is so robust that the question in the literature is rarely "does offloading happen?" — it is "what does the underlying capacity look like after months or years of consistent offloading?"

The clearest natural experiment came from spatial memory. Dahmani and Bohbot's 2020 paper in Scientific Reports compared people who navigated most often via turn-by-turn GPS with those who used it sparingly, and found that habitual GPS users performed worse on independent spatial-memory and wayfinding tasks years later. Scientific American covered the finding in plain English. Spatial memory is one of the easier capacities to study in isolation — you can hand someone a map and measure performance — and the GPS-and-memory result remains one of the cleanest demonstrations of offloading-driven skill change.

The next chapter was search. Sparrow, Liu, and Wegner's 2011 paper inScience, "Google Effects on Memory", ran four experiments showing that when participants believed information would be available later online, they remembered the information itself less reliably — and remembered the location of the information better. The paper named what most people had already noticed in their own behaviour: when retrieval is cheap, retention shifts.

The current chapter is AI. The 2024 MDPI Societiesstudy extended the cognitive-offloading framework directly to AI tools, finding an inverse relationship between AI usage frequency and self-reported critical-thinking engagement. The 2025 Frontiers in Psychology piece "Cognitive offloading or cognitive overload?" argued the question was no longer whether AI offloads — clearly it does — but whether it overloads coping in parallel. EDUCAUSE's 2025 "Paradox of AI Assistance" piece made the case for higher education specifically.

The research does not show that using external tools harms general intelligence. It does not show that calculators, GPS, search engines, or AI assistants cause cognitive decline in the clinical sense. The findings are about specific changes in how people remember, encode, recall, and engage under specific conditions — not universal claims about cognitive health.

The 2016 Risko and Gilbert review is explicit that offloading is often adaptive — using a calendar instead of holding every appointment in working memory frees capacity for tasks that benefit from it. The GPS work does not claim navigation skill loss matters for everyone equally. The Google-effects work measured shifts in retrieval, not declines in intelligence. The 2024-2026 AI work is in early stages, much of it cross-sectional and self-report-based, and should not be read as establishing causation in either direction.

What the literature consistently supports is more modest: offloading shifts where cognitive effort lands, and skills that are not exercised tend to be less practised over time. The intervention question — what, if anything, to do about it — is a separate question.

The practical question for an individual is not "should I use tools or not?" — that ship has sailed for any modern professional. The practical question is "which thinking acts do I still want to be able to do myself, and how do I keep that capability warm?"

Senwitt is one answer to the second half of that question: a daily Set across six Skills that gets at the thinking acts most easily offloaded to AI — writing, math, code, memory, reading, and reasoning. We do not claim this prevents cognitive change. We claim, narrowly, that practising things is how you stay in practice.

The design echoes a much older idea in skill literature — Ericsson's deliberate-practice tradition — applied to the modern problem of how much thinking is now done by the tools at our elbow.