Approach-avoidance bias modification with phone movements: a cross or circle on each picture says whether to pull the phone toward oneself or push it away, and the symbol is paired with the picture's category.
Configure parameters and run an interactive preview exactly as participants will experience it. No data is recorded.
Adjust parameters below, then start the preview on the right.
Include practice trials
Shown with feedback before the main task
Task parameters
BetaConfirmed on an iPhone; Android is still being checked.
Phones only: the task reads the phone's accelerometer. The web preview simulates the movements with buttons or the ↑/↓ keys.
Participants should hold the phone with both hands, ideally standing.
The result stores reaction time, response force and validity per trial. As in Zech et al. (2020), a trial is valid with a movement in the required direction, an RT of at least 200 ms and a peak of at least 1 m/s²; participants with less than 80% valid trials are flagged.
Single-session test-retest reliability is low: ICC < .3 for one session, ICCk = .73 over 8 sessions for faces (Zech et al., 2023). The task suits repeated measurement.
Approach-avoidance bias modification (Blomberg et al., 2025): a small cross (pull) or circle (push) on each picture, paired with the picture's category. A research tool, not a treatment: in a randomized trial in depression it did not change symptoms at the end of training more than sham (Blomberg et al., 2025). Measure the bias with the Approach-Avoidance Task (phone movement) in separate sessions.
Active: every picture of the trained category has a circle (push), every other picture a cross (pull). Sham: each category is pulled and pushed equally often (Blomberg et al., 2025; Wiers et al., 2011). Use study groups to assign participants to active or sham.
Default the first category (e.g. alcohol: Wiers et al., 2011; Morgenstern et al., 2026). In the active condition, Alcoholic drinks are always pushed and Soft drinks always pulled.
20–400, half per category. Default 200 (Blomberg et al., 2025).
0 = no break. Default 20: Blomberg et al. (2025) ran the 200 trials in 10 sets of 20 (whether they paused between sets is not stated).
With feedback (✓ / ✗), same rule. Default 8 (Blomberg et al., 2025; they used different pictures for practice).
One block of 200 trials after 8 practice trials (about 8 min). Blomberg et al. (2025) ran 6 sessions within 2 weeks, at most one per day, and measured the bias in separate assessment sessions: add assessment questions before and after the training sessions.
Pictures
Fixed: the acceleration perpendicular to the screen, corrected for gravity, requested at 100 Hz (Zech et al., 2020). A movement ends the trial once it exceeds +4 m/s² toward the participant (pull) or -6.5 m/s² away (push), and the trace runs until 1 s after it (Zech app). Each trace is then scored as in the preprocessing script of Zech et al.: the first acceleration peak gives the direction and the force, the RT is the last moment before it with less than 0.5 m/s². Practice feedback and the timeout clock stay 1500 ms (Zech app). Zech app = the task's published source code (osf.io/y5b32).
Outputs: the share of movements that followed the symbol; median RT and force per category and direction; the share of valid trials; the condition; and, if stored, the acceleration trace per trial. No bias score: training trials are not an assessment.
This is a researcher preview. No participant data is recorded.
Simulated participant view
Approach-avoidance bias modification with phone movements: a cross or circle on each picture says whether to pull the phone toward oneself or push it away, and the symbol is paired with the picture's category.
No data is recorded
Phones only, with the same movements, sensor and scoring as the Approach-Avoidance Task (phone movement). After a black dot for 1.5 s, a picture appears with a small cross or circle near its top: the participant pulls the phone toward themselves for a cross and pushes it away for a circle. In the active condition every picture of the trained category has a circle (pushed away) and every other picture a cross (pulled); in the sham condition each category is pulled and pushed equally often. One block of 200 trials with a break after every 20, after 8 practice trials with feedback. The web preview simulates the movements with buttons or the arrow keys.
For research on approach-bias modification, e.g. alcohol, food or emotional faces, with a sham control and separate assessment sessions, as in the smartphone trials of Blomberg et al. (2025).
Not a treatment: in randomized trials, smartphone approach-avoidance training did not change depression symptoms at the end of training more than sham, and did not change anorexia nervosa outcomes (Blomberg et al., 2025). Not an assessment: measure the bias with the Approach-Avoidance Task (phone movement) in separate sessions. Phones only. Beta: confirmed on an iPhone; Android is still being checked.
The defaults follow the smartphone training of Blomberg et al. (2025): 200 trials in sets of 20, 8 practice trials, the symbol rule (cross = pull, circle = push), 100% contingency in the active condition and 50/50 in the sham condition. Blomberg et al. ran 6 sessions of about 20 min within 2 weeks, at most one per day, and assessed the bias before and after in separate sessions; schedule the sessions and add Approach-Avoidance Task (phone movement) questions for the assessments. Assign participants to active or sham with study groups. Upload a picture set with two categories, or use the built-in drink pictures (alcohol pushed away by default, as Wiers et al., 2011, and Morgenstern et al., 2026).
The outputs describe how the participant did the training (did they follow the symbols, how fast and how forcefully), not their approach bias. Training effects are tested by comparing the active and sham groups on assessment sessions and clinical outcomes. Feedback during training is not stated in the smartphone trials; here only the practice trials give feedback.
Add Approach-Avoidance Training (phone movement) to your ESM study on SMAAT — free to get started.