Perceptual learning for peripheral attentive motion tracking transfers to the untrained hemifield and related tasks

Abstract

Purpose : Patients with central vision loss rely on peripheral vision for daily living activities such as navigation. In this study we tested whether attentive motion tracking, an essential component of navigation, could be improved in peripheral vision by perceptual learning (PL). In addition, we investigated whether any PL effects transferred to the untrained hemifield. Finally, we also tested whether PL of attentive motion tracking transferred to two related tasks: global motion perception (GMP) and static numerosity judgment (SNJ, attention), and one unrelated task: global form perception (GFP).

Methods : Participants (n = 12, Age: 24 ± 4.5y) completed 5 consecutive days of 1h PL sessions. Psychophysical threshold measurements were made for both hemifields (±10° eccentricity) at baseline and day 5 and included multiple object tracking (MOT) speed threshold (8 elements, starting speed 4°/s, partial report procedure), motion coherence threshold (GMP; 2AFC direction discrimination task for up/down random-dot-kinematograms [100 dots, 6^o/s dot speed]), form coherence threshold (GFP; 2AFC Glass pattern discrimination), and presentation time threshold for the numerosity task (SNJ; static frames from the MOT task with 8 ± 1 elements presented temporally in 2AFC). All thresholds were measured using 2-up-1-down adaptive staircases. Subjects were trained on the MOT task in a randomly chosen hemifield over 300 trials per session. PL sessions used each participant’s MOT speed threshold on day 1 and speed increased by 5% from the previous day for each subsequent session. Central fixation was ensured using an eye-tracker.

Results : Following PL, all participants improved their MOT task performance in both the trained (45 ± 10.5%, t₁₁ = 17.4, p < 0.001) and untrained (27.6 ± 12%, t₁₁ = 9.5, p < 0.001) hemifields. Improvement was also observed for GMP [trained (27 ± 13%, t₁₁ = 3.65, p = 0.004), untrained (22.5 ± 9%, t₁₁ = 3.13, p = 0.009)] and SNJ [trained (30 ± 14%, t₁₁ = 3.27, p = 0.007), untrained (27 ± 12%, t₁₁ = 2.56, p = 0.026)] tasks performance, but not for GFP [trained (t₁₁ = 1.66, p = 0.324), untrained (t₁₁ = 1.48, p = 0.393)].

Conclusions : PL improves attentive motion tracking in the peripheral vision of healthy observers. Learning partially transfers to the untrained hemifield and untrained tasks that involve motion perception and static attention.

This is a 2020 ARVO Annual Meeting abstract.