Module 01
Cognitive Psychology
Tools for the processes students are usually asked to take on trust: how attention selects and what that costs, how memory reconstructs rather than replays, and how perception and judgement are shaped by a system working with limited capacity and incomplete information.
12 tools published
Tools in this module
-
Posner Spatial Cueing
A spatial-cueing experiment the learner performs on themselves: central fixation, a peripheral cue that brightens one of two boxes, and a target that is sometimes where the cue was. Valid, neutral and invalid trials are summarised separately, so the benefit of a valid cue can be told apart from the cost of an invalid one. A seeded worked example reaches every conclusion without performing the task, and the page states repeatedly that eye position is not measured.
-
Visual Search Laboratory
Learners search original geometric displays for a bar that is tilted and solid. In the feature condition it is the only tilted item; in the conjunction condition half the distractors are tilted and half are solid, so neither property alone identifies it. Set size, target presence, search slopes and target-absent trials are all reported, alongside an untimed panel for looking at displays and a seeded worked example that reaches every conclusion without performing the task.
-
Inattentional Blindness
A demanding counting task on a dense static display that also contains a large object nobody was asked about. The learner commits to the count, submits it, answers a recognition question, and only then sees the same display again with everything labelled. Four deterministic variants let a lecturer repeat the demonstration with a different design rather than re-testing the same people, and a demonstrator route reveals the whole thing without spending a variant.
-
Change Blindness
Two versions of an original vector street scene alternate, with or without a blank between them. The learner searches for the one difference, says when they have it, and identifies what changed. The same scene can then be re-run with the blank switched off, which is the comparison the tool is built around: the picture is identical, and only the blank has changed.
-
Attentional Blink
A rapid stream of consonants contains a digit and, on most trials, the letter X. The learner predicts how second-target accuracy will depend on the gap between the two, runs four slow practice trials and then a block of 25 scored trials, and reads the resulting lag function. Second-target accuracy is reported both over all trials and conditioned on the first target being correct, so the size of that analytic choice is visible. A seeded, simulated worked example reaches every conclusion on the page without running a stream at all.
-
Stroop Interference Laboratory
An original colour-word interference task. The learner predicts where the neutral condition will sit, runs practice trials and then a scored block naming ink colours, and reads a summary that splits the usual single Stroop effect into interference above the neutral baseline and facilitation below it. A trial-mix setting turns the proportion-congruency manipulation into something the learner can run rather than be told about, and a transfer challenge asks what four further changes to the task would do to the effect.
-
Dichotic Listening and Selection Theories
A visual analogue of the competing-message paradigm, with an optional stereo tone layer synthesised in the browser. Two word streams run side by side; the learner monitors one and ignores the other, and one item in the ignored stream is planted. Two questions after each trial keep apart a physical judgement (did anything look different?) and identification of meaning (which word was it?). The attended-task rule is the load manipulation, and a seeded simulated class dataset supplies the pattern that one person's four trials cannot.
-
Dual-Task and Limited Capacity Laboratory
Two simple choice tasks - a letter judged vowel or consonant, a number judged odd or even - performed alone and then together. The three blocks run in a fixed order so that every dual-task cost is measured against the learner's own single-task baseline. Two settings turn the laboratory into an argument: making the number task harder without changing its response, and removing its per-item choice response altogether. The results panel reports the cost to each task separately, so an asymmetric cost is visible rather than averaged away.
-
Working-Memory Load Laboratory
The learner holds two, four or six items across a fixed retention interval, makes three quick judgements while holding them, and then answers one recognition probe. Material (letters or grid positions) crosses with concurrent task (syllable judgement, symmetry judgement or nothing), so a same-domain pair can be compared with a different-domain pair of identical pacing. The tool reports a load function for each run and accumulates runs in a comparison table, and it says throughout that it operationalises a demand rather than measuring anybody's capacity.
-
False Memory and Source Monitoring
The learner studies three original themed lists of eight items, each item labelled with one of two named sources, and is then tested on twelve words: six studied, three related words that were never presented, and three unrelated words that were never presented. Recognition, confidence and source attribution are collected as three separate judgements and reported as three separate measures, because the places where they come apart are the finding. A transfer challenge asks what changing source distinctiveness, delay, forewarning and list length would do to each measure.
-
Decision Framing Laboratory
Four fictional decisions about stock at risk - seedlings, servers, grain and hire bicycles - each offering a safe option and a gamble with identical expected outcomes. Two are written as gains and two as losses, assigned at random, and the equivalence is withheld until all four decisions are made. The learner can set how certain the safe option is and how the stock is described, and the reveal prints each decision beside the wording that was not shown. A seeded simulated class supplies the aggregate pattern that four decisions cannot, and the challenge asks the learner to rewrite a scenario so the framing matters less.
-
Human Attention versus AI Attention
Three stages on one sentence. The learner first resolves an ambiguous pronoun in two sentences that differ in three words, and notices that the resolution was fast and reportable. They then compute scaled dot-product attention weights over the same sentence by choosing a position, a head, a softmax temperature and whether later positions are masked. A sorting challenge over eight statements closes the loop: two are true of both, two of neither, and those four mark exactly where the analogy holds and where it fails.
What this module covers
These are the topics the module was planned around. The tools above cover them; the list is kept so the intended scope stays visible and so a contributor proposing a new tool can see what is already here.
- Selective attention and its costs What goes unnoticed while attention is directed elsewhere, and why that cost is a property of the task and the situation rather than a shortcoming of the person.
- Working memory capacity Holding and manipulating information while it decays or is displaced, and what a capacity estimate does and does not predict about anything else.
- Reconstructive memory How recall is assembled at retrieval rather than played back, and how confidence and accuracy come apart in ways that matter for eyewitness evidence.
- Perception as inference Why the same sensory input supports more than one interpretation, and what the system does to settle on one.
- Judgement under uncertainty Heuristics treated as a reasonable response to limited information and limited time, rather than as a list of faults.
A note on tasks that measure performance
Several tools in this module will ask students to do something — recall a list, respond to a display, hold a span — and then show them their own result. Those results are demonstrations of a paradigm run on a laptop in uncontrolled conditions. They are not measurements of anybody's attention, memory or ability, and the tools will say so on the page.
Every timed or performance-based tool will also offer an untimed route to the same idea, so that no student has to complete a speeded task in order to take part.
Building one of these
All five topics above are open. If you teach cognitive psychology and have a demonstration you already run by hand, turning it into a page here makes it reusable by everyone else. The contributing guide covers the folder layout, the shared interactive shell and the accessibility checks a tool needs to pass before it is merged.