Rodents' ability to segregate and recognize objects is still debated today. Our group showed that rats can distinguish figures on a background by focusing on luminance-based first-order cues. In a follow-up study, we demonstrated that rats can then learn to differentiate between figures defined by second-order texture-based cues. The aim of this current study was to design and test a novel discrimination task for mice to test whether they can differentiate between complex motion trajectories based upon second-order cues. 8 mice separated in two groups performed discrimination tasks in automated touch screen operant chambers for 28 testing days. Group 1 had an expanding white square on a black background as a target, group 2 a contracting one (Phase 1), which were then showed simultaneously during testing on the two touchscreens. In a second phase, both black and white squares on inverted backgrounds were randomly interleaved (Phase 2). For Phase 1, 3 out of the 4 mice in the contracting group reached threshold performance of 75% correct after 18.8±5.0 sessions. Interestingly, only 1 out of the 4 mice in the expanding group reached threshold level in 28 sessions. Of interest, mice in the expanding group were thrice as likely to fail to perform 30 trials per session (31.3±3.7% vs 9.8±5.5% of the time). This suggests a possible selection bias toward contracting targets, which will be further evaluated. The mice that passed threshold in step 1 had an average discrimination score of 82.8±4.6% during their first session when introduced to Step 2, indicating that they can generalize. Their score then all slightly dropped, before reaching back threshold in about 4 sessions. Our data indicate that mice can distinguish complex motion trajectories while ignoring luminance-based cues. Radially expanding/contracting targets are thus an interesting tool for probing figure-ground segregation capabilities of rodents.