When Uncertainty Matters The Selection of Rapid GoalDirected

When Uncertainty Matters: The Selection of Rapid, Goal-Directed Movements J. Trommershäuser, L. T. Maloney , M. S. Landy, Psychology and Neural Science, New York University : -500 MEGa. Move* Model for Movement under Risk 2 penalty conditions: 0 and -500 points = 4. 83 mm The choice of motor strategy depends on penalty: 100 x Mean movement end points with more complex configurations. exp. , penalty = 0 exp. , penalty = 500 model, penalty = 500 S 5 x (mm) Results: S 1 penalty: 500 S 2 x (mm) 60 30 0 -30 y x y • the reward structure of the environment x y x x, y: mean movement end point [mm] y [mm] • the mover's own motor variability. Trommershäuser, Maloney, Landy (2003) JOSA A, in press. Trommershäuser, Maloney, Landy (2003) Spatial Vision, 16, 255 -275. Maloney, Trommershäuser, Landy (2003) VSS S 4 90 points per trial Consequence: S 3 x (mm) Experiment 2 MEGa. Move Model: The Expected Gain Surface Key assumption: 100 The mover chooses the motor strategy that maximizes the expected gain , taking into account motor uncertainty. 5 “practiced movers” 1 session: 12 warm-up trials, 6 x 2 x 16 trials per session, 24 data points per condition In this simple task, a motor Expected Gain Surface strategy is characterized by the distribution of end points around a mean end points 10 Each mean end point corresponds per trial 5 to an expected gain. This expected 90 60 -0 gain depends only on the subject’s 30 motor uncertainty and the rewards 0 -5 -30 and penalties present in the -10 <-60 environment. -10 -5 0 5 10 15 20 target: 100 x (mm) We can compute, for any subject, penalty: -500 = 4. 83 mm an expected gain surface. penalty: 0 S 2 2 penalty conditions: 0 and -500 points S 3 <-60 5 “practiced movers” 1 session: 12 warm-up trials, 6 x 2 x 16 trials per session, 24 data points per condition y (mm) The red target is hit: -500 points S 1 y (mm) 100 4 S 4 x (mm) y (mm) 100 3 R = 9 mm y (mm) Kassi Price, 2001 US Nationals 2 Results: 3070 points x (mm) = 4. 83 mm 700 ms time limit. -500 1 -32 points x (mm) The green target is hit: +100 points L : 100 points (2. 5 ¢) -32 points The Experimental Task Speeded movement: . . Our results indicate that humans take both costs and their own movement uncertainty into account in planning movement. : -500 100 points -400 points y (mm) Here we report the outcomes of two experimental tests of the model in simple environments where there are explicit gains and losses associated with the outcomes of movements. : 100 points (2. 5 ¢) y (mm) The MEGa. Move* Model predicts that subjects will take account of their own motor uncertainty in planning movements. Hit targets, with Fingertip, avoid penalties. Experiment 1 Mean movement end points with stimulus configurations at different orientations. MEGa. Move Model: Effect of Motor Uncertainty y (mm) Movement under Risk y (mm) VSS 2003 TALK Sarasota, FL x (mm) x exp. , penalty = 0 exp. , penalty = 500 model, penalty = 500 S 5 x (mm) *: Maximum Expected Gain Model of Movement planning x [mm] = 4. 83 mm x [mm] Supported by NIH EY 08266 and HFSP RG 0109/1999 -B, J. T. funded by the DFG (Emmy-Noether Programm)