sortN

Experiments

sort1-sort6 define the CORRECT ORDER of the stimuli in a CFPT drag-to-reorder trial (a single horizontal sortable list, historically “Cambridge Face Perception Test” style). They are consumed only when the test column of the trial equals the exact string CFPT. Each sortN cell must contain one of the values that also appears in the trial’s stim1-stim6 cells; the sequence sort1, sort2, ... sortN is the target arrangement the participant is supposed to produce by dragging.

They are NOT drop-zone/category labels, and they are unrelated to the survey-style responseType: sort feature (that one is scored against the key column, and never reads sortN).

Syntax

Copy the trial’s stimulus values into the sort columns in the correct final order: sort1 holds the stimulus that belongs first, sort2 the one that belongs second, and so on.

• Each cell must be a character-for-character copy of one of the trial’s stim1 to stim6 values as written in the CSV. For image trials that is the filename exactly as typed in the stim columns (if the stim cell says faceA.jpg, the sort cell must say faceA.jpg too); for word trials, the word exactly as written.
• Matching is exact and case-sensitive; there is no fuzzy matching and no extension handling.
• There are as many sort columns as stim columns, but scoring always assumes exactly 6 items (see Tips & gotchas).

Options

Value	Meaning
exact copy of one of the trial’s stimN values	This position (N of sortN) is where that stimulus belongs in the correct final arrangement.
empty / anything not matching a stim value	No parsing error, but both the recorded final order and the error score are silently corrupted for the affected stimuli. There is no validation; matching is the only semantics.

There is no keyword syntax, no separators, no numeric mode: the value is used only for exact-match lookups against the stim values.

Defaults & missing values

• An absent sortN column counts as empty cells: nothing matches, so a CFPT trial’s recorded order and score are corrupted. Fill all six sort cells on every CFPT row.
• On non-CFPT trials the values are completely ignored, but they are echoed back into the results file like all recognized columns.
• test=CFPT2 displays a sortable image grid but never reads sortN and never scores; the trial is saved like a normal button response.

Works with

• test: must be exactly CFPT (case-sensitive) to activate the sort columns, and only on single-screen trials.
• stim1 to stim6: supply both the displayed items and the identifiers matched against sortN. The initial display order is always the stim column order; stimOptions: random does NOT shuffle a CFPT list.
• presTime: drives a visible countdown (the value in ms is shown as seconds). Under 10 seconds the timer turns red; at 0 the list is disabled and dimmed with “Time up!” but the trial does NOT end on its own; the participant must still press the button. The usual presTime auto-advance does not apply to CFPT trials, and an empty presTime yields a timer that never expires.
• button1: pressing the button finalizes the arrangement and saves the trial. On CFPT trials the single button is repositioned near the bottom of the screen.
• leftArrow / rightArrow: their texts are displayed beside decorative arrow images on CFPT trials.
• type / feedback: data is recorded only when type is test and no feedback screen is active. A CFPT trial with type = practice records nothing.
• Recorded output: response is the final arrangement as the matched sort values joined by ;; correct is the total displacement, the sum over stimuli of the absolute difference between correct position and final position (0 = perfect arrangement, larger = worse).

Examples

Image ordering, 6 faces morphed away from a target, participant must drag into similarity order within 60 s:

type	test	stimFormat	stim1	stim2	stim3	stim4	stim5	stim6	sort1	sort2	sort3	sort4	sort5	sort6	presTime	button1
test	CFPT	.jpg	faceD	faceA	faceF	faceB	faceE	faceC	faceA	faceB	faceC	faceD	faceE	faceF	60000	Done

Displayed left-to-right as faceD..faceC (stim order); correct arrangement is faceA..faceF (sort order). Clicking “Done” records response like faceA;faceC;faceB;faceD;faceE;faceF and correct = total displacement (here 2 if only faceB/faceC are swapped).

Word ordering (words trial, no target, plain text stims):

type	test	stim1	stim2	stim3	stim4	stim5	stim6	sort1	sort2	sort3	sort4	sort5	sort6	presTime	button1
test	CFPT	third	first	fifth	second	sixth	fourth	first	second	third	fourth	fifth	sixth	30000	Submit

Real-world naming pattern (modeled on the library CFPT identity-sort tests in the corpus). In the real files each face appears as one identifier plus a numeric morph-distance suffix; stim1-stim6 hold a scrambled presentation order and sort1-sort6 hold the same six files in ascending morph distance (the intended similarity ordering). The sort* values below are taken verbatim from a real file; the stim* order is an illustrative scramble of that same set:

type	test	target	stim1	stim2	stim3	stim4	stim5	stim6	sort1	sort2	sort3	sort4	sort5	sort6	iti
test	CFPT	Iden_02	Iden_02_36	Iden_02_12	Iden_02_72	Iden_02_24	Iden_02_60	Iden_02_48	Iden_02_12	Iden_02_24	Iden_02_36	Iden_02_48	Iden_02_60	Iden_02_72	500

Tips & gotchas

Caution

Scoring assumes exactly 6 stimuli. With fewer than 6, unfilled positions can carry stale entries from the previous CFPT trial, contaminating both the recorded final order and the error score; with more than 6, items beyond the sixth are displayed but never scored. Build CFPT trials with exactly 6 stimuli.

• The correct value is a total rank displacement, not right/wrong. Treat it as an error magnitude when analyzing CFPT rows; anything that reads it as a 0/1 flag will misread them.
• If any sortN value fails to match a stim value exactly, the recorded final order silently contains missing entries, with no warning.
• The sort columns are not category drop zones and there is no all-or-nothing scoring; they define the single correct ordering that the error score is measured against.