Reconstruction & Attribution outline
Memory is reconstructed from cues and current models, not replayed; what you recall reflects the predictions active now, not a faithful record of the past.
Memory isn't playback---it's reconstruction. When you recall an event, the brain doesn't retrieve a stored video file. It generates a prediction about what should have happened based on current schemas, active task-sets, and available cues, then fills gaps with whatever fits. This is why memory feels vivid and confident even when it's wrong: the predictive processing system produces a coherent story, and you experience that story as memory. The reconstruction is guided by the same machinery that generates predictions about the present---it's just aimed backwards instead of forwards.
This architecture explains why your memory of an event can shift over time. As your current frames and models change, so does what you "remember." The past you reconstruct today reflects the neural pathways and priors that are active now, not necessarily what was encoded then. Source attribution---where a memory came from, whether it's real or imagined, whether it's yours or someone else's---is similarly reconstructive. Under ambiguity or pressure, the system binds details to whichever source fits the current model. This is how confabulation and false memories arise: the prediction is confident, the story is coherent, and there's no internal flag marking it as fabricated.
Attribution extends beyond memory to explanation. When you explain behaviour---yours or others'---you're running a causal model that's shaped by your current frames. Individualist frames attribute outcomes to personal traits ("they failed because they're lazy"). Structural frames attribute outcomes to context and systems ("they failed because the system set them up to fail"). The frame determines the story, and the story determines what you attend to next and which actions seem sensible. This is why ideological frames are so powerful: they don't just shape what you perceive in the moment; they shape what you remember having perceived, and how you explain it.
The practical lever is to recognise that memory and attribution are active constructions, not passive retrievals. They're tunable by priors, aesthetics, and social frames. If you want to change how someone remembers or explains something, you don't argue with the content directly---you change the cues, the context, or the model they're using to reconstruct. Design environments and practices that make the intended story the most coherent fit, and the reconstruction will follow.
How can you think with this?
These heuristics help you apply this neural system:
Ways to think with this
Practical ways to use this neural mechanism in understanding behaviour
WIP: Memory is prediction aimed backwards
Memory isn't playback---it's reconstruction. The brain generates a prediction about what should have happened based on current schemas, active task-sets, and available cues, then fills gaps with whatever fits. This is the same predictive processing machinery that operates in the present, just aimed backwards. The past you reconstruct today reflects the models and pathways active now, not necessarily what was encoded then.
So what can you do? Recognise that memory is tunable by current frames and models. If you want to change how you or someone else remembers something, don't argue with the content---change the cues, the context, or the model being used to reconstruct. Design environments and practices that make the intended story the most coherent fit, and the reconstruction will follow.
WIP: Frames determine explanations
Attribution---how you explain behaviour, yours or others'---is shaped by the causal frames you're using. Individualist frames attribute outcomes to personal traits. Structural frames attribute outcomes to context and systems. The frame determines the story, and the story determines what you attend to and which actions seem sensible. These frames are ideologies in the functional sense: they filter what gets bound together in memory and explanation.
So what can you do? Understand that explanations reflect the frame being used, not objective truth. If you want to shift how someone explains or remembers, shift the frame. Change which features get attended to, which patterns get highlighted, which causal links are made salient. The reconstruction follows the frame, so changing the frame changes the story.
WIP: Confidence doesn't track accuracy
The predictive processing system produces coherent reconstructions, and you experience that coherence as confidence. This is why false memories feel vivid and real---the prediction is confident, the story is coherent, and there's no internal flag marking it as fabricated. Confidence tracks the fit between reconstruction and current models, not the match between reconstruction and past events.
So what can you do? Don't trust confidence as a signal of accuracy. Vivid, detailed memories can be wrong because they're reconstructions that fit current models well, not recordings of the past. When accuracy matters, corroborate with external evidence. And when you're trying to update someone's memory, focus on changing the models that drive reconstruction, not on arguing about details they're confident about.
Sources
- neurotypica/content/archive/anatomy-memory.md
- analects/repressed-memories.md