Invariants and priors in tactile perception of object motion

Transcription

1 Haptics Symposium 2016: Workshop Haptic Invariance Invariants and priors in tactile perception of object motion Alessandro Moscatelli Universität Bielefeld - Università di Roma Tor Vergata Universität Bielefeld

2 Motion Invariants and Spatial Constancy V prop Invariants and priors in tactile motion perception 2

3 Motion Invariants and Spatial Constancy V prop V tact Invariants and priors in tactile motion perception 2

4 Motion Invariants and Spatial Constancy v prop = v tact Invariants and priors in tactile motion perception 3

5 Motion Invariants and Spatial Constancy v prop = v tact v surf = v prop + v tact Invariants and priors in tactile motion perception 3

6 Motion Invariants and Spatial Constancy v prop = v tact v surf = v prop + v tact Spatial constancy refers to the capacity of the perceptual system of estimating accurately the velocity of an external surface, v surf, in a world-framed coordinates. Invariants and priors in tactile motion perception 3

7 Motion Invariants and Spatial Constancy v prop = v tact v surf = v prop + v tact Spatial constancy refers to the capacity of the perceptual system of estimating accurately the velocity of an external surface, v surf, in a world-framed coordinates. Human tactile system partially lacks of spatial constancy: ˆv surf = 04v prop + v tact Invariants and priors in tactile motion perception 3

8 Experimental Paradigm Invariants and priors in tactile motion perception 4

9 Experimental Paradigm 2AFC Experiment: Invariants and priors in tactile motion perception 4

10 Experimental Paradigm 2AFC Experiment: K (reference) Vs T (comparison) Invariants and priors in tactile motion perception 4

11 Experimental Paradigm 2AFC Experiment: K (reference) Vs T (comparison) T (reference) Vs K (comparison) Invariants and priors in tactile motion perception 4

12 Results A 1.0 B 20 Probability FASTER 0.5 PSE [mm/sec] Δ v [mm/sec] KT Condition TK Invariants and priors in tactile motion perception 5

13 Three Hypotheses 1. The gain of the two signals is intrinsically not unitary Invariants and priors in tactile motion perception 6

14 Three Hypotheses 1. The gain of the two signals is intrinsically not unitary 2. Perceived tactile speed is proportional to the spatial frequency of the patterned stimulus. Invariants and priors in tactile motion perception 6

15 Three Hypotheses 1. The gain of the two signals is intrinsically not unitary 2. Perceived tactile speed is proportional to the spatial frequency of the patterned stimulus. 3. The difference in precision between the two signals generates the illusion (Bayesian hypothesis). Invariants and priors in tactile motion perception 6

16 The Stationarity Prior Model Probability Finger motion Posterior (K) K Prior Likelihood P K σ K σ P σ K - ve 0 + ve + Probability T T σ T σ T P σ P Tactile motion (T) - ve 0 + ve = Probability K +T σ K +T - ve 0 + ve Velocity Invariants and priors in tactile motion perception 7

17 Experimental Paradigm Same task as in Experiment 1. Participants (N = 12) where divided in three groups, using each one of the following: 1. A belt having an homogeneous, smooth surface 2. A belt having a low-frequency, textured surface (ξ = 0.07 mm 1 ) 3. A belt having a high-frequency, textured surface (ξ = 0.3 mm 1 ) Invariants and priors in tactile motion perception 8

18 Experimental Paradigm Same task as in Experiment 1. Participants (N = 12) where divided in three groups, using each one of the following: 1. A belt having an homogeneous, smooth surface 2. A belt having a low-frequency, textured surface (ξ = 0.07 mm 1 ) 3. A belt having a high-frequency, textured surface (ξ = 0.3 mm 1 ) We modeled the response as follows: Φ 1 [P (Y ij = 1)] = β 1 v ± β 2 ξ Invariants and priors in tactile motion perception 8

19 Results: PSE 20 PSE [mm/sec] 0 20 KT Smooth TK KT Low Freq. TK KT High Freq. TK Invariants and priors in tactile motion perception 9

20 Results: Noise 4 Bias [cm/sec] ξ smooth low high ΔJND [mm/sec] Invariants and priors in tactile motion perception 10

21 Mechanism A B ξ Tactile Noise ξ Tactile Noise Bias Bias Invariants and priors in tactile motion perception 11

22 Experimental Paradigm Unimodal, tactile discrimination task. Invariants and priors in tactile motion perception 12

23 Experimental Paradigm Unimodal, tactile discrimination task. The belt consisted of two distinct surfaces, having either a smooth or a high-frequency texture (ξ = 0.25 mm 1 ) Invariants and priors in tactile motion perception 12

24 Experimental Paradigm Unimodal, tactile discrimination task. The belt consisted of two distinct surfaces, having either a smooth or a high-frequency texture (ξ = 0.25 mm 1 ) In two intervals, participants (N = 8) contacted either the smooth (S) or the rough (R) textured part of the belt and reported in which interval the belt moved faster. Invariants and priors in tactile motion perception 12

25 Results PSE [mm/sec] SR Condition RS Invariants and priors in tactile motion perception 13

26 Results 20 JND [mm/sec] SS Condition RR Invariants and priors in tactile motion perception 14

27 Results PSE [mm/sec] KT 40 TK KT TK SR Exp 1 Exp 2 Exp 3 RS Invariants and priors in tactile motion perception 15

28 Summary The study showed a weak spatial constancy in touch. Invariants and priors in tactile motion perception 16

29 Summary The study showed a weak spatial constancy in touch. The spatial frequency of the stimulus and the perceptual noise affect the perceived tactile velocity Invariants and priors in tactile motion perception 16

30 Summary The study showed a weak spatial constancy in touch. The spatial frequency of the stimulus and the perceptual noise affect the perceived tactile velocity The perceptual bias is reduced if participants are exposed to a large range of tactile speeds: Sensory calibration? Invariants and priors in tactile motion perception 16

31 Summary The study showed a weak spatial constancy in touch. The spatial frequency of the stimulus and the perceptual noise affect the perceived tactile velocity The perceptual bias is reduced if participants are exposed to a large range of tactile speeds: Sensory calibration? ˆv = fλ Invariants and priors in tactile motion perception 16

32 References Moscatelli, A., Hayward, V., Wexler, M., Ernst, M. O. (2015). Illusory Tactile Motion Perception: An Analog of the Visual Filehne Illusion. Scientific Reports, 5, Moscatelli, A., Scotto di Cesare, C., Ernst, M. O. (In Prep.). A novel illusion provides insight in motion processing in touch. Dallmann, C. J., Ernst, M. O., Moscatelli, A. (2015). The role of vibration in tactile speed perception. Journal of Neurophysiology, 114(6), Invariants and priors in tactile motion perception 17

33 Thank you for your attention Invariants and priors in tactile motion perception 18

34 Spatial Constancy in the Tactile System A tracking B judgement of perceived movement v prop + v tact = v surf v prop = v surf v tact = 0 50 mm trial time Invariants and priors in tactile motion perception 19

35 Spatial Constancy in the Tactile System Moscatelli et al (2015) Invariants and priors in tactile motion perception 19

36 Spatial Constancy in the Tactile System a probability away from observer b LI 10 Main Control MO Main Control V surf [mm/s] PSE [mm/s] Main Control Invariants and priors in tactile motion perception 19

37 Spatial Constancy in the Tactile System The bias in motion direction is caused by a different gain between tactile and hand velocity Invariants and priors in tactile motion perception 19

38 Spatial Constancy in the Tactile System The bias in motion direction is caused by a different gain between tactile and hand velocity ˆv surf = v prop + v tact Invariants and priors in tactile motion perception 19

39 Spatial Constancy in the Tactile System The bias in motion direction is caused by a different gain between tactile and hand velocity ˆv surf = 04v prop + v tact Invariants and priors in tactile motion perception 19

40 Spatial Constancy in the Tactile System The bias in motion direction is caused by a different gain between tactile and hand velocity ˆv surf = 04v prop + v tact We run a second experiment to further test this hypothesis Invariants and priors in tactile motion perception 19

41 Vibration as Cue to Tactile Speed A B Acceleration (m/s 2 ) 1 s Surface speed (cm/s) Acceleration (m/s 2 ) 1 s Frequency (Hz) 300 Acceleration (m/s 2 ) Acceleration (m/s 2 ) Surface speed (cm/s) Frequency (Hz) Invariants and priors in tactile motion perception 20

42 ˆv K = β 1 v ˆv T = β 1 v + β 2 ξ Invariants and priors in tactile motion perception 21

43 ˆ v = ˆv K = β 1 v ˆv T = β 1 v + β 2 ξ { ˆv T ˆv K = β 1 v + β 2 ξ if KT, ˆv K ˆv T = β 1 v β 2 ξ if T K. Invariants and priors in tactile motion perception 21

44 ˆ v = ˆv K = β 1 v ˆv T = β 1 v + β 2 ξ { ˆv T ˆv K = β 1 v + β 2 ξ if KT, ˆv K ˆv T = β 1 v β 2 ξ if T K. P (Y j = 1) = P ( v ˆ > 0) = Φ( v ˆ > 0) = Φ(β 1 v ± β 2 f), Invariants and priors in tactile motion perception 21

45 ˆ v = ˆv K = β 1 v ˆv T = β 1 v + β 2 ξ { ˆv T ˆv K = β 1 v + β 2 ξ if KT, ˆv K ˆv T = β 1 v β 2 ξ if T K. P (Y j = 1) = P ( v ˆ > 0) = Φ( v ˆ > 0) = Φ(β 1 v ± β 2 f), Φ 1 [P (Y ij = 1)] = β 1 v ± β 2 ξ. Invariants and priors in tactile motion perception 21

46 Results: Model Fit 1.0 ag dk ik mas 0.5 Probability FASTER ks mh rr yg kn mn rj sm Cond. ξ KT TK smooth low high V [mm/sec] Invariants and priors in tactile motion perception 22