S.M.M. Field Manual

1. Situation · The Terrain

Behavioral science distinguishes two learning engines, and manipulation uses both. Classical (Pavlovian) conditioning teaches that one stimulus predicts another — Pavlov’s dogs salivated at a bell that predicted food (Pavlov, 1927) [C5]; the Rescorla–Wagner refinement is that learning tracks surprise, not mere co-occurrence (Rescorla & Wagner, 1972) [C4]. So neutral cues acquire the charge of what they precede: a notification sound becomes arousing, a logo inherits positive affect, a partner’s tone triggers dread or longing before a word is processed. Operant (instrumental) conditioning teaches that one’s own behavior produces consequences — Thorndike’s Law of Effect (1911) [C5], formalized by Skinner into reinforcement and punishment (Skinner, 1938) [C5]. Control when and how often reward arrives and you control how durable and compulsive the behavior becomes.

Why unpredictable reward is most powerful

The single most important fact: how predictably a reward follows behavior matters more than its size (Ferster & Skinner, 1957) [C5]. Continuous reward builds fast but extinguishes fast; variable-ratio schedules — reward after an unpredictable number of actions — produce the highest, steadiest rates and the greatest resistance to extinction. This is the slot-machine schedule, deliberately engineered into feeds, matches, loot boxes, and trading apps (T18.1 ★, T18.5, T18.7). Skinner’s “superstition” pigeons developed rituals around accidentally-timed food (Skinner, 1948) [C4]; humans do the analogous refresh-tap rituals around any variable reward.

The brain rewards surprise; wanting ≠ liking

Midbrain dopamine encodes a reward prediction error — a fully predicted reward produces little response; an unexpected one produces a large one (Schultz, Dayan & Montague, 1997; Schultz, 2016) [C4]. A variable schedule maximizes prediction error, and thus the motivating signal (T18.21). Crucially, dopamine tracks wanting (incentive salience), dissociable from liking (Berridge & Robinson, 1998) [C4]: conditioning can sensitize wanting while liking flattens — the mechanistic signature of compulsion. “I don’t even enjoy it anymore” is a coherent description of a heavily conditioned habit. Repetition in a stable context then produces a habit — cue-triggered performance independent of current value (Wood & Rünger, 2016) [C4], the cue → routine → reward loop (Duhigg, 2012) [C3], with automaticity plateauing after a median ~66 days (Lally et al., 2010) [C3].

Table 6.1 — Reinforcement schedules and their manipulative use

2. Enemy Forces · The Loop, Shaping & Predation

Design disciplines build products around the habit loop explicitly: Fogg’s model holds a behavior occurs when Motivation, Ability, and a Trigger converge (Fogg, 2009) [C3]; Eyal’s “Hooked” chains Trigger → Action → Variable Reward → Investment, where the final investment step raises switching costs and loads the next trigger (Eyal, 2014) [C3]. These are ethically neutral tools; Category 18 turns them against the user via cue engineering (notifications T18.2, “we miss you” T18.11), friction manipulation (infinite scroll T18.6, autoplay T18.12 remove stopping cues while exit friction is added), and reward & investment (variable-schedule likes T18.4, streaks T18.3 that convert quitting into a loss).

Shaping reinforces successive approximations (Skinner, 1938) [C5]: a manipulator rewards a small action, then a slightly larger one, so no single step is large enough to trigger refusal — the conditioning under foot-in-the-door (T6.1) and salami escalation (T6.9, T6.13). The gamification stack (T18.14–T18.18) is shaping formalized. The defensive key: compare every step to the original baseline, not the previous step. The purest predatory form is machine gambling (Schüll, 2012) [C3]: the near-miss recruits win-related reward circuitry despite being an objective loss (Clark et al., 2009) [C3], and loss-chasing escalates stakes. Loot boxes (T18.7) import variable-ratio, near-miss, and completion pressure into games sold to minors.

Table 6.2 — The manipulative habit loop, stage by stage

Table 6.3 — Conditioning techniques by domain

Relational conditioning (high-harm, recognition level)

Intermittent reinforcement in relationships (T18.9 → T7.19; T22.11) is the variable-ratio schedule enacted with affection, approval, and safety as the reward. Delivered unpredictably — interspersed with coldness or withdrawal — it produces the same extinction-resistant persistence a slot machine does, and the attachment strengthens precisely because the reward is unpredictable. This is the conditioning core of trauma bonding and coercive control: the intermittency forges the bond. Treated at recognition level only — the diagnostic is structural (unpredictable oscillation; wellbeing routed through regaining another’s warm state), and interrupting it typically requires stable outside support and often professional help.

Conditioning is not inherently manipulative — it is the mechanism of all learning, and used honestly builds skill, exercise, and saving habits. The marks of ethical conditioning: the reward is proportionate and real; cue and friction serve the user's own goals (easy to start and easy to stop); and the design survives disclosure. Manipulative conditioning engineers compulsion the user would not choose and depends on their not noticing a loop was installed. Because the loop is built from cue, schedule, and friction, the defensive posture is architectural.