Both drills force retrieval before feedback. You locate the evidence — or find the target — under your own control, and only then is your answer evaluated. In a classic experiment, students who were tested on material retained substantially more of it a week later than students who simply restudied it, an advantage now known as the testing effect (Roediger & Karpicke, 2006). The effort of reconstructing an answer is what consolidates it.

Time limits and unrevealed answers are deliberate. Conditions that make practice harder in the moment — generation, spacing, reduced cues — tend to produce more durable learning than fluent, easy study, a principle Robert and Elizabeth Bjork (2011) call desirable difficulties. The fifteen-second flashlight scan compresses this effect into a single drill.

Comprehension research models reading as two layers: a textbase of propositions and a situation model of what the text is about (Kintsch, 1998). Evidence matching exercises the second layer, demanding that a question be connected to the sentence that licenses its answer. How precise that connection can be depends on lexical quality — how exactly a reader knows words and their paraphrase relations (Perfetti, 2007) — and on cohesion, the web of lexical ties that holds a text together (Halliday & Hasan, 1976). The feedback notes on this site are written to make those ties visible.

The flashlight drill targets the perceptual layer of reading. Eye-movement research shows skilled readers fixate selectively and skip predictable words (Rayner, 1998), while visual-search studies show that distinctive features — digits, capitalised names, rare word forms — can be detected almost effortlessly (Treisman & Gelade, 1980; Wolfe, 1994). Automating word recognition frees attention for meaning (LaBerge & Samuels, 1974), which is what fast scanning trains.