We take the ability to write for granted. It is actually quite a complex non-linear activity driven by our goals and purpose. To understand this process and how AI can support it while you maintain ownership, it’s helpful to understand the cognitive activities of writing.
Several cognitive models explain the writing process, and the core of most models is based on Flower and Hayes’ cognitive activities of writing (1981). The model identifies the mental operations that underlie writing as distinct activities: planning, translation (text production), and reviewing.
Their framework also recognizes that writing does not occur in a vacuum. The task environment, including the audience, topic, purpose, and text produced during the process, influences the writer’s decisions.
In this article, I’ll explain each phase of the Flower and Hayes cognitive model and how to use AI as a thinking partner during the process in a way that maintains your cognitive ownership and agency.
The Cognitive Model of Writing
Their framework consists of three mental processes of writing: planning, translating (producing text), and reviewing. If you do a lot of writing, you are probably aware of how fluidly you shift between the three processes, looping back to earlier stages as you work. Here is how the researchers explain each process.
The Planning Process
During the planning process, writers form an internal representation of knowledge that may be represented in different symbol systems, such as images or keywords, to represent a network of knowledge. Planning consists of three critical sub-processes:
- Generating: This involves retrieving relevant information from long-term memory, including external sources. Writers use retrieval cues to bring forth knowledge networks, which may be well-organized thoughts or fragmented.
- Organizing: This sub-process involves giving structure to ideas by identifying categories and subcategories of concepts. It may involve planning the start and the end of a piece.
- Goal-setting: Writers decide what they want to accomplish and set sub-goals, such as how to write and what to say.
The Translation Process
Translating is the act of verbalizing abstract ideas and thoughts into language. It is often described in terms of several interacting activities, though they are not considered subprocesses. These are listed below.
- Managing simultaneous constraints, like genre conventions and audience expectations.
- Relying on automaticity for sentence construction, word choice, and similar operations.
- Transforming ideas and reorganizing knowledge during expression.
The Reviewing Process
This activity involves two subprocesses: evaluating and revising. Reviewing occurs when a writer reads their writing to further their thinking or to evaluate what they have written. It can interrupt any process at any point.
- Evaluating: Judging the quality of the text or plans against the established goals.
- Revising: Modifying text or unwritten thoughts based on evaluation.
Using AI Tools for Collaboration
If you’ve noticed a backward slide in your writing or critical thinking skills due to reliance on AI tools, recent research may help you overcome this trend. The idea is to do your thinking first and use AI as a collaborator second. For example, when writing, identify the purpose, generate the initial ideas, make the key decisions, and write in your own words. Use AI to extend, challenge, or refine your thinking. Then you’re doing the effortful work that lets you keep building skills, to intentionally maintain ownership, and exercise agency over your work.
Below are examples of how to use AI tools during each cognitive activity of writing. The instructions and prompts illustrate ways to challenge your thinking rather than replace it. Provide the appropriate context when using them, and adjust as needed. This approach ensures conscious control over the level and type of AI support you use during the writing process.
Using AI During the Planning Process of Writing
| Mental Process/Activity | Examples of Strategic Use of AI |
|---|---|
| Generating: Retrieving relevant information from long-term memory and external sources. | Generate initial ideas independently. Then use AI for research (tools like Consensus, Elicit, NotebookLM) and to identify missing perspectives or gaps in your evidence. Evaluate all summaries before you adopt them. Example prompts: • What perspectives am I missing? • Which arguments have the strongest evidence? Which have the weakest? • What assumptions am I making about the topic? |
| Organizing: Structuring the ideas. |
Think through the structure of your writing in an outline or diagram. Then provide the LLM with purpose, audience, context, and pedagogy. Example prompts: • What is another way to structure these concepts? • Where might my audience get confused or misinterpret the message? • Where is my logic failing? |
| Goal-setting: Identifying what you want to accomplish and setting mini-targets. | Define your goal and sub-goals independently. Then use AI to challenge your direction as in the examples below. Example prompts: • Where is there a mismatch between my outline and my goals? • Which sub-goals might lead me in the wrong direction? • From the perspective of the target audience, where does this goal fail to address their needs or expectations? |
Using AI During the Translating Process of Writing (Producing Text)
| Mental Process/Activity | Examples of Strategic Use of AI |
|---|---|
| Translating: Converting ideas into language while managing constraints. | Draft your text independently. Then use AI to evaluate how you express your ideas. Focus on clarity, tone, and language. Use this feedback to revise in your own words. Example prompts: • Where might the language be unclear or misinterpreted by the audience? • Where is the writing too dense or difficult to follow? • Where does the writing sound dull or clichéd? |
Using AI During the Reviewing Process of Writing
| Mental Process/Activity | Examples of Strategic Use of AI |
|---|---|
| Evaluating: Judging the quality of the text against goals and criteria. | Review your draft against your goals first. Then use AI to evaluate how well the piece achieves its purpose. Use this feedback to guide your revisions, not to replace them. Example prompts: • How well does my draft achieve its intended purpose for the audience? • What key points will the learner/reader take away from this? |
| Revising: Modifying text or unwritten thoughts based on evaluation. | Identify revisions independently. Then use AI to help prioritize which changes will have the greatest impact. Use this guidance to revise the text in your own words. Example prompts: • What are the highest-impact revisions I can make to achieve my goal? • Where does my current draft fall short of the intended outcome? • What content could be removed without weakening the overall purpose? |
Conclusion
The goal is not to avoid AI support, but to maintain your role as the thinker, decision-maker, and writer. AI can help you explore ideas, challenge your reasoning, and refine your language. However, your cognitive ownership diminishes when a tool or another person defines the substance of your work. See Writing with AI for Instructional Design for more tips.
References
- Chandrasekaran, M., Koh, J. Y., & Feng, S. (2022). The effect of AI writing assistance on writing quality. In Proceedings of the Workshop on Intelligent and Interactive Writing Assistants (In2Writing 2022) (pp. 8–13). Association for Computational Linguistics.
- Flower, L. & Hayes, J. (1981). A cognitive process theory of writing. College composition and communication 32, 4 (1981), 365–387.
- Gerlich, M. (2024). AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies, 15(1), 6; https://doi.org/10.3390/soc15010006.
- Hardman, P. (April 16, 2026). The Cognitive Offloading Paradox. Dr. Phil’s Newsletter. https://drphilippahardman.substack.com/p/the-cognitive-offloading-paradox
- Reza, M., Thomas-Mitchell, J., Dushniku, P., Laundry, N., Williams, J. J., & Kuzminykh, A. (2025). Co-writing with AI, on human terms: Aligning research with user demands across the writing process. Proceedings of the ACM on Human-Computer Interaction. Advance online publication.
[Disclaimer: The content in this RSS feed is automatically fetched from external sources. All trademarks, images, and opinions belong to their respective owners. We are not responsible for the accuracy or reliability of third-party content.]
Source link
