Shoot

Part of Deep Processing, the parent dimension for encoding work.

Shoot is reading a source to answer questions written before it was opened, with every answer going onto a map of your own that is allowed to keep changing. It is the second step of the Bear Hunter System — Aim wrote the questions, and Shoot spends the source answering them, not collecting from it. That is the point where a source becomes useful: the curiosity Aim built gets satisfied, and a working map takes shape while the learning is still happening.

The step

The questions sit beside the source, and the reading is done to answer them. Each thing the source offers gets one of four treatments: it answers a question; it goes under the chunk it belongs to; it adds a relation, if it changes the map; or it gets marked incomplete. Source order is followed only where it is the best structure on offer — the sequence belongs to the reader, which is Order Control running inside the step.

Watched from across a desk, a session going well looks like a stretch read, a stop, a flip back to something earlier, a few lines written, a pause to join them to what is already there, and the source re-entered for the question that just appeared. Back and forth, not front to back. The order that costs is the prescribed one: front to back asks for a passage’s meaning, its place, and its relevance all at once, and that is where retention drops. The visible disorder is the load being spent in one place at a time.

In a lecture or live video the step runs compressed: short lines rather than sentences, the map growing as the session runs, detail hung on chunks that already exist, question marks on the unresolved, no transcript mode. Five to ten minutes afterwards tidies the rough map — a working figure of this system, not a finding. Where the pace cannot be paused at all: backbone plus open questions, with Shoot finished afterwards. On a text the rhythm is a pause at each section — which question this part answered, where the answer lands, whether a new chunk appeared. A highlight earns its place the same way, by answering a question or supporting a chunk; without that processing it changes the page and leaves the map unchanged.

What counts as an answer

An answer in this step is logical and interconnected, and both halves carry weight. Logical means its origin is visible: why it has to be that way can be seen, which turns it into something to reconstruct instead of something to hold. Interconnected means it arrives attached to what it relates to, and the number of attachments is the point — what connects to many things gets kept, what connects to nothing gets dropped.

Relations come first because of what they do to the facts. Work that only answers what is this leaves each fact standing alone, the kind the mind discards. With relations in place, facts start arriving as consequences of them — the difference between a detail that can be derived and one that has to be memorized. The calibration shows in anything learned outside a curriculum: a violinist can say roughly how long the bow is without ever learning the number, because the knowledge sits inside a structure that produces it. Some material still has to be memorized outright; the work is keeping that set small rather than pretending it is empty.

The map is not decoration on top of this: across 55 studies, building structures of labelled relations beat conventional study by about half a standard deviation.

The map stays provisional

The map is the product, and it is never finished while the reading runs. Concepts move, chunks get renamed, an overloaded area splits, thin areas merge, missing relations get drawn in, and details that stopped mattering come off. A map still matching its first guess at the end of a session is the thing to notice, because understanding that improved would have moved something. This is also why the source’s order was never the default: a structure borrowed whole is a map that cannot move.

What waits, and what does not

Detail waits until the structure it belongs to exists. Encoded early it costs more than it would to miss: met before its surroundings, it sits isolated — harder to retain, headed for memorization instead of inference. Met after the concept layer exists, the same detail tends to stop needing to be held; what would not stick alone follows from the logic around it. Layers of Learning owns the theory behind that ordering.

The exception is immediate: a detail that clarifies a relation gets captured on the spot, and so does one that heads off a misunderstanding about a chunk. Past the exception, a detail earns its place by doing work — explaining a relation, supporting a chunk, clarifying an example, preparing for assessment, heading off a likely forgetting, exposing a gap. Appearing in the source is not on the list.

What slips past is not lost: missed details surface in later retrieval, in the cut-down pass, or on the next pass through the source — the system runs in layers, each catching what the one before missed. Not every question gets a clean answer either, and the reaching is still where the learning happens, because the question drives the search for relations; one too advanced for this pass gets marked and left.

A session that worked reads like this:

• The main Aim questions have answers on the map.
• Each detail kept is attached to a chunk it serves.
• The map no longer matches the first guess, and it does not follow the source’s order.
• The main relations can be said out loud with the source closed.
• Unresolved questions carry a visible mark.

Where it goes wrong, and what it costs

Trouble in this step usually reads upstream. When the questions were good, Shoot is the easy half: direction exists, the work is answering, details have somewhere to go, the map moves, weak chunks show themselves. When it is hard, either the questions were thin or the old linear pass took over as the source opened. The signs sit on the page — a map mirroring the source’s table of contents, full sentences accumulating, chunks that never move because moving them feels like wasting the drawing, every detail at equal weight, linear blocks with arrows added afterwards, scattered work with nothing holding it together. Catching any of this mid-session is Self-Regulation‘s work, the control layer watching while the session runs.

The session shows	The move
Full sentences accumulating	One Aim question gets answered from memory, source closed.
Detail piling up	Each detail names the chunk it serves; one with no chunk parks or goes.
A chaotic map	The map rebuilds from its three to five largest chunks.
Relations that stay vague	The connecting question runs on one pair at a time: how this relates to that.
An Aim that was too thin	The session pauses and re-Aims before more reading happens.

Continuous writing is the standing temptation: it feels productive, and it removes the pressure to process — deciding what a passage amounts to is the friction doing the work, so notes carry conclusions and transcription stays in the source. What the note research supports is narrower than the way it gets quoted: verbatim capture is the weak form, and typing makes verbatim drift easy. The non-linear form itself takes practice, and a failed first attempt is usually flawed form rather than a weak method — Non-Linear Note-Making owns the full treatment. Collecting text is the cheap move here, the one The Shortcut Problem names: the session gets discharged while the thing it was for gets skipped. A Shoot can also look correct and still be copying — right format, moved boxes, nothing processed — the case The Technique Is Only as Good as the Thinking It Produces covers.

The price is front-loaded: slower per pass than transcription, with the feeling of coverage withheld — nothing looks finished until the relations hold, so certainty arrives late. That cost falls as the technique stops needing attention, and what the fast pass buys instead is a sense of completion that does not track what was built. The boundary sits at material with no relational structure to find — conventions, arbitrary curriculum facts, vocabulary. Those belong in memorization tools, and forcing them onto the map is one of the ways a map gets ruined.

Effort during the step is the expected condition — the work is supposed to feel like figuring something out, and Cognitive Load & What Mental Effort Is Trying to Cue reads what that effort signals. The opposite signal is just as usable: study easy enough to produce drowsiness, when sleep is not the cause, is a technique reading — the processing has stopped. The quit signal is one specific test. Main relations that still cannot be said from memory at the end of a session put the fault in the questions, and the next move is re-Aiming rather than shooting harder.

The loop

With practice the two steps stop being two: answering produces the next question, and it gets aimed and shot without a seam. The cycle cannot be forced — it grows out of competence in each step separately, and anyone not yet cycling keeps running them apart. Many of a topic’s questions do not exist until answering starts; that is why the loop exists at all. What a session leaves is a working map — questions answered, relations clear, details attached, the unresolved marked — ready for Skin, the pass that cuts it down to what can be rebuilt without the source.

Open Questions

• Where the line sits between a detail that clarifies a relation and one that is merely near it — the hardest judgment the step asks for.

Related

• Deep Processing — the parent dimension this step belongs to.
• Bear Hunter System — the loop this is the second step of.
• Aim — where the questions come from, and the step to return to when this one scatters.
• Skin — the next step: the map cut down to what can be rebuilt without the source.
• Non-Linear Note-Making — the note form this step writes in, and why delayed writing does the processing.
• Layers of Learning — the layer theory behind detail waiting for structure.
• Order Control — the sequence belonging to the reader rather than the source.
• Self-Regulation — the control layer that notices the warning signs while the session runs.
• Cognitive Load & What Mental Effort Is Trying to Cue — what the effort in this step signals, and what overwhelm means when the questions were thin or detail came early.
• The Technique Is Only as Good as the Thinking It Produces — a Shoot that looks correct and is still copying.
• The Shortcut Problem — collecting text as the cheap move the step displaces.

Sources

• Dunlosky, Rawson, Marsh, Nathan & Willingham (2013), Psychological Science in the Public Interest 14(1) — elaborative interrogation and self-explanation rated moderate utility; the support for reading to answer pre-written why/how questions.
• Pressley, McDaniel, Turnure, Wood & Ahmad (1987), Journal of Experimental Psychology: General — the original elaborative-interrogation demonstration.
• Chi, de Leeuw, Chiu & LaVancher (1994) — self-explanation: explaining a passage in terms of what it relates to.
• Nesbit & Adesope (2006), Review of Educational Research — meta-analysis of 55 studies; structures of labelled relations versus conventional study, d ≈ 0.46.
• Pollock, Chandler & Sweller (2002) — isolated-elements effect; Ausubel (1960) — advance organizers. Detail met before a frame exists is isolated and expensive.
• Mueller & Oppenheimer (2014) — verbatim capture as the weak form; Morehead, Dunlosky & Rawson (2019) and Urry et al. (2021) — the longhand-versus-laptop format claim did not reproduce.