working memory
In plain wordsThe small amount of information held in mind right now while doing a task. For an AI model, that is roughly whatever currently sits in its context window.
- 1
Agents
The information available to an AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done. for its current decision: in practice the contents of the model's context windowcontext window, n.: How much an AI model can read and keep in view at once: the instructions, the conversation so far and any attached documents. Text that does not fit has to be cut, summarized or left out. — instructions, recent conversation, retrieved passages, tool results and any scratchpadscratchpad, n.: A space where an AI writes rough working — calculations, plans, notes — before or while producing its answer, much as a person jots on scrap paper. — together with state variables the harnessharness, n.: The software around an AI model that gives it instructions and tools, carries out its actions and decides when it stops. The model is the engine; the harness is the rest of the car. tracks for the current step. In the CoALA framework it is the hub through which the agent's reasoning, long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later. retrieval and actions communicate. Like human working memory it is small relative to everything the agent could access, so context engineeringcontext engineering, n.: Managing an AI model's reading material: choosing, trimming and arranging the instructions, documents, tool results and memories it is given on every turn, so that it has what the task needs and little else. is largely the management of what occupies it, using retrieval, compactioncompaction, n.: Like replacing a stack of rough notes with a one-page digest: when an AI's conversation grows too long for its memory, the history is condensed into a summary and the work continues from there. and offloading to files to keep it relevant.
Keep only the current subtask's files in working memory and leave the rest on disk.
- 2
Cognitive science
The mind's small, short-lived workspace, in which a few items are kept active and worked on during tasks such as mental arithmetic, following a conversation or planning. Alan Baddeley and Graham Hitch's influential model divides it into a central executive, a phonological loop and a visuospatial sketchpad, later adding an episodic buffer; capacity estimates range from about seven items (George Miller, 1956) down to about four chunks.
Origin
The phrase appears in George Miller, Eugene Galanter and Karl Pribram's 'Plans and the Structure of Behavior' (1960); Alan Baddeley and Graham Hitch's 1974 model made it central to cognitive psychology. AI cognitive architectures such as Soar and ACT-R built in a working memory, and the CoALA framework (2023) applied the concept to LLM agents.
Usage note
Agent builders often use 'short-term memory' as a synonym; psychologists distinguish short-term memory (passive storage) from working memory (storage plus manipulation).
Related
- context windowcontext window, n.: How much an AI model can read and keep in view at once: the instructions, the conversation so far and any attached documents. Text that does not fit has to be cut, summarized or left out.
- long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later.
- agent memoryagent memory, n.: The ways an AI assistant or agent keeps information — your preferences, past conversations, lessons from earlier tasks — and brings it back later, beyond what fits in a single conversation.
- scratchpadscratchpad, n.: A space where an AI writes rough working — calculations, plans, notes — before or while producing its answer, much as a person jots on scrap paper.
- context engineeringcontext engineering, n.: Managing an AI model's reading material: choosing, trimming and arranging the instructions, documents, tool results and memories it is given on every turn, so that it has what the task needs and little else.
- cognitive architecturecognitive architecture, n.: A blueprint for a complete thinking system — how memory, attention, learning and decision-making fit together — used both to build general AI agents and to model how the human mind works.
- long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later.
Sources
- Baddeley, A. D. & Hitch, G. (1974). Working memory. In The Psychology of Learning and Motivation, vol. 8. Academic Press.
- Sumers, T. R., Yao, S., Narasimhan, K. & Griffiths, T. L. (2023). Cognitive Architectures for Language Agents. arXiv:2309.02427. arxiv.org/abs/2309.02427v3
Mentioned in
agent memoryagent memory, n.: The ways an AI assistant or agent keeps information — your preferences, past conversations, lessons from earlier tasks — and brings it back later, beyond what fits in a single conversation., long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later., scratchpadscratchpad, n.: A space where an AI writes rough working — calculations, plans, notes — before or while producing its answer, much as a person jots on scrap paper., episodic memoryepisodic memory, n.: Memory of particular past events — what happened, when and in what order. AI agents keep records of earlier conversations or attempts so they can recall them and learn from them.