Visual Statistical Learning - Fisher & Aslin

Technical Manual

Script Author: Katja Borchert, Ph.D. (katjab@millisecond.com), Millisecond

Millisecond LLC thanks Turk-Browne et al (2005) for providing the 12 original shapes used in this study!

Last Modified: January 31, 2023 by K. Borchert (katjab@millisecond.com), Millisecond

Script Copyright © Millisecond Software, LLC

Background

The script "Visual Statistical Learning I" adapts the procedure used by Fiser & Aslin (2002) to study the unprompted statistical learning of temporal relationships inherent in a visual presentation of moving shapes:

References

Fiser, J. & Aslin, R.N. (2002). Statistical Learning of Higher-Order Temporal Structure from Visual Shape Sequences. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28, 458-467.

Duration

10 minutes

Description

Participants are presented with a 5-6min "movie", in which 12 basic shapes moved one after the other across the monitor from, seemingly "changing" from one shape into another when they cross a black bar in the center of the screen. The "change", however, is not random as the 12 shapes were grouped into 4 triplets with strict rules regarding to which shape can follow another, e.g. triplets have to stay together and have to be presented always in the same order (A->B->C). To test whether participants have learned the inherent temporal shape relationships, the movie-triplets (=familiar) are pinned against "novel" triplets in a balanced forced-choice task, in which participants are asked to decide which of the two triplet pairs are more familiar to them.

Procedure

PART I: Shape Sequence "Movie":
This script implements a simple imitation of the original movie used by Fiser & Aslin (2002).
In Fiser and Aslin's study, participants watched a shape appear from under a black bar that was positioned
in the middle of the screen. The shape then moved to the left of the screen, changed direction and moved back
towards the bar, behind which it disappeared once more. When it "reappeared", the shape had "morphed"
into the next shape, which stayed on the same trajectory and moved to the right side of the screen,
changed direction and moved back towards and behind the bar again (and so on). Each shape was presented for 1s.

The underlying shape sequence in this script meets the same criteria used by Fiser & Aslin (2002):
(1) 12 shapes divided into 4 triplets
(2) Triplets stay together
(3) Each triplet is presented 24 times (=288 shape presentations)
(4) No immediate repeats of triplets (e.g. ABCABC)
(5) No repeat of triplet pairs (e.g. ABCDEFABCDEF)
(6) Each triplet appears the same number of times in each third of the sequence (8times)

In addition, this script offers the opportunity to randomly
(1) select from a pool of 10 possible sequences that all meet these criteria. (=> vary sequence across participants)
(2) assign the 12 shapes to the 4 triplet pairs (=> triplet pairs are not constant across participants)

PART II: Forced-Choice
In the Forced-Choice part of the experiment, the script pins the 4 familiar triplet pairs
(ABC, DEF, GHI, JKL - from the sequence) against 4 novel triplets (AEI, DHL, GKC, JBF)*.
These novel triplets are made up of shapes that had never directly followed each other during the
sequence presentation - changes to these triplets can be made under trial.assignments (e.g. change triplets to
partial triplets as used in Fiser & Aslin (2002), Experiment2).

Participants get to view 2 sequences of 3 shapes. Each shape is presented for default=500ms**
with a break of 1s (s. Fiser & Aslin).
Shapes appear in a left- right-left sequence (s. Fiser & Aslin). Participants are asked to decide
which one is more familiar.
(default TASK: press "1" for first, press "2" for second -> change keys to press under values).
Each of the 16 triplet-pairings is presented twice, once with the familiar sequence being presented
first and once with the familiar sequence being presented second. The resulting 32 possible
forced-choice tasks are selected at random.

•Fiser & Aslin (2002) do not specify the novel triplet combinations used.
The triplet combinations used in this script are taked from Turk-Browne et al (2005).
The automaticity of visual statistical learning. Journal of Experimental Psychology: General, 134, 552-564.

Fiser & Aslin (2002) Do Not Specify The Screen Duration Of The Individual Shapes In The Forced-Choice Task.

Stimuli

Millisecond LLC thanks Turk-Browne et al (2005) for providing the 12 original shapes used in this study!
See section Editable Stimuli

Instructions

Instructions are not original to the task. They are provided by Millisecond
as htm/html pages and simple page elements and can be edited either by changing
the provided htm/html files or directly under Editable Instructions.
To edit htm/html-files: open the respective documents in simple Text Editors such as TextEdit (Mac)
or Notepad (Windows).

Summary Data

File Name: vsl_fiser_summary*.iqdat

Data Fields

NameDescription
inquisit.version Inquisit version number
computer.platform Device platform: win | mac |ios | android
startDate Date the session was run
startTime Time the session was run
subjectId Participant ID
groupId Group number
sessionId Session number
elapsedTime Session duration in ms
completed 0 = Test was not completed
1 = Test was completed
propCorrect Proportion correct familiar triplet identifications
meanRTCorrect Mean latency (in ms) of correct familiar triplet identifications
sdCorrect Standard deviation (in ms) of correct familiar triplet identification latencies

Raw Data

File Name: vsl_fiser_raw*.iqdat

Data Fields

NameDescription
build Inquisit version number
computer.platform Device platform: win | mac |ios | android
date Date the session was run
time Time the session was run
subject Participant ID
group Group number
session Session number
blockCode Name of the current block
blockNum Number of the current block
trialCode Name of the current trial
trialNum Number of the current trial
list.tripletSequence.currentIndex The index of the selected tripletsequence (e.g. the third sequence in the list)
selectedLetter The letter presented in the shape presentation sequence
inde The letter position in the shape presentation sequence
a-l The indices of the shapes (from the item list shapes) that are assigned to shapes a-l
novel11-novel43 The indices of the shapes (from the item list shapes) that are assigned to novel triplets
pairing Determines which familiar and which novel triplets were compared and in which order
(e.g. 15 -> triplet pair ABC against novel triplet 1, familiar first)
15 = old ABC triplet, novel1 triplet
51 = novel1 triplet, old ABC triplet
...
48 = old JKL triplet, novel4 triplet
84 = novel4 triplet, old JKL triplet
shape1-shape6 The shapes presented during the forced-choice trials with shape 1 being the first
response The SCANCODE of the key-response of the participant
2 (scancode) => number '1' (first sequence more familiar)
3 (scancode) => number '2' (second sequence more familiar)
responseText The character code of the key pressed
correct Whether this is a correct response (1= correct; 0 = error)
latency How long the participant took to make response (in ms); measured from onset of task question

Parameters

The procedure can be adjusted by setting the following parameters.

NameDescriptionDefault
waitForcedChoice Determines the pretrialpause before the first shape triplets are shown in the forced-
choice task (default is 1s)
leftKey The left key used to respond if the first sequence is more familiar
(default = "1")
rightKey The right key used to respond if the second sequence is more familiar
(default = "2")