zeyus / touchtracker Goto Github PK

mousetrap csv compatible flutter (Android, iOS, Web) touch tracking implementation of Falk, Winter, and Spivey's “Candle-Candle-Candle-Candy: Continuous Attraction Toward Previously Seen Phonological Competitors.” experiment

License: Open Software License 3.0

Kotlin 0.14% Swift 0.44% Objective-C 0.04% Dart 93.92% HTML 4.25% R 1.22%

cognitive-science flutter psychology psychology-experiments experiments experiment research

touchtracker's Introduction

touchtracker

Work in progress

Flutter project for Mobile + Web.

Like mousetracking, but for touch. Output CSV / logs are compatible with the mousetrap R library.

Example data analysis can be seen in the example_data_analysis dirctory of this repo.

Tested working on Android and Web.

Current state

Try it out

You can run the latest touchtracker web release to give it a go.

Future plans

After finishing using the app for a research project, I want to decouple the experiment and make it somewhat easy to use the app as an experiment builder (or maybe just a "player") which can be customized with various assets (sounds, images, text prompts). Using the Candle-Candy experiment as a demo / tutorial.

License

OSL 3.0

touchtracker's People

Contributors

Stargazers

Watchers

touchtracker's Issues

Block draggable widget from leaving viewport

check coordinates of dragable and don't allow positioin ondrag to go outside the window / screen

Investigate using flame instead of straight flutter

https://pub.dev/packages/flame

Fix race condition and inconsistent Flutter draggable callback firing

initially I thought it was because of issue #5, but that's not the case, on closer inspection, there's a problem with Flutter not consistently calling onDragEnd, onDragComplete or onDragCancelled...
Sometimes they fire, sometimes they don't, and it's not clear why though I suspect it's because I'm making draggable and dragtarget do something they weren't made for.

This can be handled within the controller by checking if the callback has been fired once for the drag end/cancel/complete event.

Change target detection method if possible

this shouldn't be a big issue but who knows now...

currently the drag indicator is slightly bigger than a finger, so it has lead to occasional dropping of the Draggable inside the dragtarget but not quite half way so it doesn't trigger the events...

the behaviour / target event should fire as soon as the edge of the indicator touches the dragtarget box

Stimuli class should throw exception on identical pair

Shouldn't have ever gotten that far 😂

align csv to mousetrap / opensesame

The bare minimum needed is the columns required by the mousetrap R package import http://pascalkieslich.github.io/mousetrap/reference/mt_import_mousetrap.html

mt_import_mousetrap(
  raw_data,
  xpos_label = "xpos",
  ypos_label = "ypos",
  timestamps_label = "timestamps",
  mt_id_label = NULL,
  split = ",",
  duplicates = "remove_first",
  unordered = "warn",
  reset_timestamps = TRUE,
  digits = NULL,
  verbose = FALSE
)

Sensible default suggested mapping for starters:

Csv column	Details
log_sequence	sequential id
timestamp	normal timestamp
experiment	experiment name
practice	1 or 0
experiment_start_time	timestamp
experiment_end_time	timestamp
experiment_elapsed_time	microseconds
subject_id	subject id
condition	overarching condition
target_stimulus	stimulitarget.a or b, (pair target)
target_position	a or b (l / r)
non_target_stimulus	stimulitarget.a or b, (pair non-target)
non_target_position	a or b (l / r)
stimulus_a_times_displayed	number of times stim displayed
stimulus_b_times_displayed	number of times stim displayed
trial_type	condition descriptor
trial_subtype	additional condition descriptor
trial_sequence	trial id 0+
trial_start_time	timestamp
trial_end_time	timestamp
trial_elapsed_time	microseconds
correct	1 or 0
response_time	microseconds
response_time_average	calculate average
velocity_average	calculate average
distance_step_average	calculate average
distance_total	calculate total
distance_direct	calculate total from start to finish point
timestamps_tracking	[ t1, t2, ... , tn ] in microseconds
xpos_tracking	[ x1, x2, ... , xn ] tracking x coordinate
ypos_tracking	[ y1, y2, ... , yn ] tracking y coordinate
velocity_tracking	[ v1, v2, ... , vn ] tracking step velocity
distance_tracking	[ d1, d2, ... , dn ] tracking step distance
angle_tracking	[ θ1, θ2, ... , θn ] tracking step angle
fullscreen	1 or 0 (hopefully always true on mobile)
viewport_height	screen height
viewport_width	screen width
device_dpi	screen dpi
device_info	all available deviceinfo

replace audioplayers with just_audio

Sadly audioplayers has met its end with this project it seems that there's no way to work around its inconsistent start timing and patchy callback firing...

the start timing issue occurs even when every audio file is precached using AudioCache, the timing issue is absolutely critical because if you want to use this for an experiment, each trial should load in the same amount of time and right now it can be from instant to 10 seconds per trial which feels very jarring while being a subject and breaks up any motor rythym.

https://pub.dev/packages/just_audio

just_audio should also allow calculation of duration on all supported platforms so we don't need to rely on a clunky future.delayed