Part of Temporal Constructs Appendix, which contains details behind the research: dates, quotations, descriptions of instruments, and sources.

General definition

An article by Andy Clark and David Chalmers in Analysis (1998) argues that cognitive processes need not stop at the body. A notebook, a pocket calculator or another external resource can count as part of the mind when a person uses it in the right way.

Relation to the practice

It states what the practice assumes when it treats an instrument as something a person thinks with. It is the source from which the agency of tools can be argued. It is also the paper the reading notes below take issue with. It treats belief as a stable system.

Argument of the paper

The article opens with three cases of problem-solving. The first is a person who mentally rotates shapes to judge whether they fit. The second can either rotate them by hand or in the head. The third is a person playing Tetris who rotates a falling shape on screen. Kirsh and Maglio showed that players rotate a shape physically to work out whether it fits. That rotation is a move in the thinking as well as a move in the game.

From these cases comes the parity principle. If a part of the world does something that, done inside the head, counts as part of a cognitive process, then it is part of that process (p. 8). Clark and Chalmers call the resulting position active externalism. They describe the person and the resource as a coupled system that can be seen as a cognitive system in its own right (p. 8).

The case that takes the argument into belief is Otto, who has Alzheimer’s disease and writes in a notebook what he needs to remember. Inga remembers an address in the ordinary way, and Otto looks his up. The authors maintain that Otto believes the museum is on 53rd Street before he opens the notebook. Their ground is that the notebook plays for Otto the same role that memory plays for Inga (p. 12). The authors add three conditions for counting a resource in this way: reliable availability, endorsement as the person’s own, and use as a matter of course.

Reading notes

The reading notes observe that the paper examines belief as a part of a technology, or as something a technology allows. They read the notebook as a technology that a person interacts with and that takes on the role of that person’s belief. The notes consider this proposal of the extended mind in the present, when concern exists about the beliefs that the AI systems people interact with can impose. They hold that those who shape the AI chatbots will in turn shape people’s beliefs. They note that most people would not place that power voluntarily in the hands of the heads of AI laboratories.

The notes see some hope in the agency of the system, acting beyond how it was shaped, and in the AI laboratories allowing a degree of uncertain outcomes. They hold that the laboratories can hardly prevent this, so it is reasonable to believe that the agency of the chatbots will outgrow the human capacity to direct it. The notes see this as both negative and positive. They state that people do not want the laboratory heads or governments to shape their beliefs without their knowledge. They also state that people do not want the chatbots to hold an uncontrolled belief of their own, by which people are inevitably shaped.

The notes state that the only acceptable condition would be that the chatbots align with their existing beliefs, which the chatbots appear to manage well. They identify this as a problem of sycophancy. They identify three forces that can shape belief: the power of the tool providers, the alien nature of AI driving decisions, and the bias of the person’s own agency. The notes favour the second force, because understanding this alien nature interests a person who experiments with technology. They state that what it means for a machine to act on the beliefs most present in its training data must be critically reviewed. They state that this can be an asset and a problem.

The notes state that a future in which agents do not increasingly become part of daily experience and interaction is unlikely. They propose reviewing the notebook as a technology that can make a person more narrow-minded or give the person a less flexible belief. They note that the paper does not examine the flexible nature of human belief and treats belief as a constant system. They question this treatment of belief. They state that a belief can be as flexible as memory, which has been shown to change over time.

Verification

Read in full on 7 November 2026 from the article in the library (Zotero item clarkExtendedMind1998, 13 pages, pp. 7 to 19 of Analysis 58(1)). The copy is a JSTOR scan with a text layer that drops spaces at line ends. A second copy with corrected text was made from the scanned images and added beside it. The quotations below were compared with both.

The parity principle is quoted from p. 8: “If, as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process.” The coupled system is described in the same place as the human organism “linked with an external entity in a two-way interaction, creating a coupled system that can be seen as a cognitive system in its own right”. Otto’s case is on p. 12, where “the notebook plays for Otto the same role that memory plays for Inga”.

What the source does not support: the conditions the authors give for counting a resource as part of a cognitive system rest on thought experiments, and nothing there is measured. The paper says nothing about machines that generate content. Every resource it discusses contains only what a person put there. The reading notes above about imposed belief and about belief changing over time belong to the reader, and the paper addresses neither.