Approach-Avoidance (AAT)

One picture at a time, after a 1.5 s fixation cross. Swiping down pulls it closer and it grows; swiping up pushes it away and it shrinks.

Useful in addiction, eating, anxiety and health-behaviour research where automatic approach tendencies toward cue pictures are a target outcome, including repeated measurement in daily life.

Not a diagnostic instrument. One short session gives a noisy bias score, so it suits designs with repeated sessions or group comparisons better than precise single-person scores.

Upload a validated picture set with two matched categories (e.g. the Amsterdam Beverage Picture Set for alcohol vs. soft drinks), or pilot with the built-in drink pictures. The defaults follow the smartphone swipe AAT of Meule, Richard, Dinic & Blechert (2019): the content rule (push one category, pull the other, then the reverse in a second block), each picture twice per block (64 trials with the built-in pictures), and practice with animal and household pictures not used in the task, as in Meule et al. (2019): 2 blocks of 10 trials before the task, also with your own image set. Keep the reversed second block so each category is both pushed and pulled. The picture-format rule remains available, but on a touchscreen a rule based on an irrelevant feature gave an unreliable bias score (Meule, Lender et al., 2019).

• Response rule: picture content (default; Meule et al., 2019) or picture format (classic AAT; Rinck & Becker, 2007)
• Pushed format or category, counterbalanced at random per participant (default; Meule et al., 2019; Rinck & Becker, 2007)
• Second block with the reversed rule (content rule; default on, Meule et al., 2019)
• Pictures: built-in drink pictures or your own image set with two categories
• Repetitions per picture and direction (default 2, Meule et al., 2019)
• Practice with animal and household pictures not used in the task, as in Meule et al. (2019): 2 blocks of 10 trials by default, or the task's own pictures; on or off. The practice rule (push animals and pull household objects, then the reverse) and the feedback after each practice response have no published value
• Response time limit (default 2000 ms, Zech et al., 2020)
• Fixed: 1500 ms fixation cross before each picture and a 200 ms minimum response time (Zech et al., 2020); zoom to 1.7× when pulled and 0.33× when pushed (Rinck & Becker, 2007); a swipe counts after 170 px or 20% of the screen height, whichever is smaller (no published value)

• Approach bias per category (median push time − median pull time)
• Bias score (category A − category B)
• Bias score from movement onset
• Wrong-direction and too-slow trials
• Swipe path per trial

A positive bias for a category means it is pulled faster than it is pushed. A positive bias score means a stronger approach tendency toward category A than toward B. Difference scores from one short session are unreliable; aggregate across sessions or model the trial-level data.

• Approach-avoidance tasks with a zooming picture have been used to measure approach tendencies toward alcohol pictures.
• Smartphone versions of the AAT have been developed for measurement outside the laboratory.
• On a touchscreen, the bias score had a split-half reliability of r = .84 when the picture content decided the direction and r = −.24 when an irrelevant feature (frame colour) did (Meule, Lender et al., 2019). A methods study of AAT data likewise found irrelevant-feature versions, on average, less valid and much less reliable (Kahveci et al., 2024).

• Meule, Richard, Dinic & Blechert (2019), a smartphone swipe AAT (source of the defaults)
• Meule, Lender, Richard, Dinic & Blechert (2019), reliability of touchscreen AATs by response rule
• Zech et al. (2020), a mobile approach-avoidance task
• Rinck & Becker (2007), the zooming approach-avoidance task
• Kahveci et al. (2024), how scoring and design choices affect AAT reliability and validity
• Wiers et al. (2009), approach tendencies toward alcohol

• Approach-Avoidance (phone movement)
• Go/No-Go
• Delay Discounting — Adjusting Amount