Learn it → Apply it → Save as Skill
The presence of other people changes how you perform, but not always for the better. An audience or a co-worker tends to boost performance on tasks you already know well, and hurt performance on tasks that are new, complex, or still being learned.
The Dr. Wallace's Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
When someone works alongside you or watches you, arousal rises and your most automatic response gets stronger. On easy, well-practiced tasks that automatic response is usually correct, so you speed up. On hard, unfamiliar tasks it usually isn't, so you slow down or make more mistakes.
Important: the direction of the effect flips with task difficulty, not with the person watching. The same audience helps and hurts the same person depending on what they're asked to do.
The model is most useful whenever work will happen in front of others, be recorded, be reviewed live, or be compared on a leaderboard.
The method is a short diagnostic, not a guarantee. It tells you which direction presence will likely push performance, so you can design around it.
Decide whether the task, for this specific person right now, is well-learned and simple or new and complex.
Identify who will be present: a co-actor doing the same task, a silent bystander, or an evaluator who will judge the result.
Simple + well-learned tends to improve with presence. Complex, new, or still-being-learned tends to suffer.
Practice new or complex work in private first. Save audiences, live reviews and leaderboards for tasks that are already mastered.
Open a use case to see what the model changes.
Routine, well-practiced work (processing tickets, data entry, familiar reports) tends to tolerate an open floor. Work that requires learning something new or solving an unfamiliar problem tends to suffer from ambient presence and should get quiet space or focus time.
A public leaderboard can speed up reps on scripted, well-rehearsed calls. The same visibility can hurt a rep who is still learning a new product line or handling an unusually complex case, where mistakes become more likely under watching eyes.
Presenting a well-rehearsed update on a video call can go smoothly. Debugging a new, unfamiliar problem live on a screen-share, in front of an audience, is exactly the complex-task condition most likely to produce visible struggle.
Reviewing or writing familiar, well-understood code with a partner watching can feel smooth and even faster. Learning an unfamiliar framework or language while someone watches over your shoulder is a harder setting for the same person.
Have learners practice a new skill privately or in low-stakes small groups first. Introduce an audience, a timer, or an evaluator only once the skill has become close to automatic.
Quality-monitoring and call recording can sharpen performance on routine scripts. For a rep still learning the job, the same monitoring can raise error rates on complicated calls until the role becomes second nature.
Social facilitation is not one sliding scale from "bad" to "good." It is the interaction of two separate factors: how much arousal the presence of others creates, and whether the task's dominant response is correct.
Rises with a co-actor doing the same task, an attentive audience, or an expected evaluation. Stays low with a blindfolded or inattentive observer.
Illustrative, not a measured score — higher means more arousing presence.
High on simple, over-learned, well-rehearsed tasks. Low on new, complex, or still-being-learned tasks, where the automatic response is often the wrong one.
Illustrative, not a measured score — higher means the automatic response is more often correct.
This pairs two classic, separately documented findings to show the simple-versus-complex split rather than inventing a single composite statistic.
Expand each item for the documented history and the honest state of replication.
Triplett noticed competitive cyclists posted faster times when racing against another rider than when racing alone, then tested the idea in children who wound fishing line onto reels, alone and side by side. Many children reeled faster paired, though some "overstimulated" children did worse. It is widely cited as the first experimental social-psychology study, though the original sample was small and not fully standardized. A 2005 statistical reanalysis by Strube found the coaction advantage was not significant between subjects in Triplett's data and only marginally significant within subjects, and a larger preregistered replication (McHugh et al., 2020s, over 400 children aged 7–13) did find children were faster paired, with the effect stronger for girls than boys. Triplett's design cannot fully separate "someone is simply present" from "I am competing against them."
Floyd Allport coined "social facilitation" in his 1924 book Social Psychology, based on his own coaction research. For decades afterward, different studies found presence sometimes helped and sometimes hurt performance, with no unifying explanation. Robert Zajonc resolved the contradiction in a 1965 Science paper: presence raises general arousal, which strengthens whatever response is "dominant," or most likely to occur. On simple, well-learned tasks the dominant response is usually correct, so performance improves. On complex or new tasks it is usually wrong, so performance suffers. Zajonc and Sales (1966) supported this with human subjects, and Zajonc, Heingartner and Herman (1969) extended it to cockroaches running simple versus complex mazes.
Two competing refinements followed. Cottrell and colleagues (1968, 1972) argued that mere presence isn't enough: the effect appeared only when observers were paying attention and could evaluate the performance, not when they were blindfolded or indifferent — this is evaluation-apprehension theory. Baron's distraction-conflict theory (1986), building on Sanders, Baron and Moore (1978), argued instead that others are simply distracting: dividing attention between the task and the person raises drive and narrows focus, which helps simple tasks and hurts attention-demanding ones. Modern work (Belletier, Normand & Huguet, 2019) suggests both routes exist side by side rather than one being the full story.
Bond and Titus's 1983 meta-analysis of 241 studies and roughly 24,000 participants, spanning 1898 to 1980, confirmed the simple-versus-complex pattern but found mere presence explained a small share of performance variance, and physiological arousal measures didn't reliably track presence across studies — a persistent weak point for drive theory specifically. A 2020 preregistered replication of Zajonc's cockroach study (Halfmann et al.) found cockroaches slowed down with others present on both simple and complex mazes, which did not fully replicate the original pattern. The broad task-complexity effect remains influential and widely taught, but treat any single precise number from an individual old study as illustrative of a pattern, not as a fixed constant.
An audience reliably helps only when the task is simple or already well-learned for that person. For new or complex work, the same audience tends to hurt.
Performance anxiety research gets more attention, but the classic finding is two-directional. Routine, rehearsed work often speeds up with others present.
Difficulty is relative to the individual's current skill, not an absolute property of the task. The same spreadsheet is simple for an expert and complex for a trainee.
A co-actor doing the same task, a passive bystander, and an evaluator who will judge you are not interchangeable; the apprehension and distraction they create differ.
Specific percentages from individual 20th-century studies describe that study's sample and task. The durable finding is the direction of the pattern, not a precise figure.
Social facilitation is about individual performance changing when others are present or co-acting. It does not describe a group's collective decision quality or reduced individual effort inside a team, which are separate, already-documented phenomena.
the concern is reduced individual effort inside a group task where contributions are hard to separate out, not a change in skill or speed from being watched.
the concern is a group's collective decisions becoming worse due to pressure toward consensus, not one person's individual task performance.
the reaction to being observed is severe, persistent distress rather than an ordinary, situational dip in complex-task performance.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply Social Facilitation, what questions to ask, what evidence to seek, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
Not strictly. Modern extensions of the research look at video calls, livestreams, and recorded sessions as presence-like conditions. The arousal and evaluation-apprehension mechanisms are thought to generalize to "being seen," though most of the classic evidence comes from in-person studies, so treat remote-work applications as a reasonable extension rather than a directly tested finding.
Yes. A presentation might combine a well-rehearsed opening (helped by the audience) with live, improvised question-and-answer (hurt by the audience) in the same ten minutes. Classify at the level of the specific sub-task, not the whole event.
Blascovich and colleagues (1999) found that when task demands roughly match a person's resources, an audience tends to produce a "challenge" cardiovascular response that supports performance; when demands exceed resources, it tends to produce a "threat" response that impairs it. This maps onto, and helps explain physiologically, the same simple-versus-complex split described by drive theory.