Neuroscience Letters, 138 (1992) 161 164 © 1992 Elsevier Scientific Publishers Ireland Ltd. All rights reserved 0304-3940/92/$ 05.00
161
NSL 08549
Interaction of vestibular and proprioceptive inputs for human self-motion perception F. Hlava6ka a, T. Mergner b and G. Schweigart b alnstitute of Normal and Pathological Physiology, Slovak Academy of Science, Bratislava (Czechoslovakia) and bNeurological University Clinic, Freiburg ( FRG ) (Received 6 December 1991; Revised version received 20 January 1992; Accepted 22 January 1992)
Key words: Human; Perception; Self-motion; Vestibular stimulation; Proprioceptive stimulation; Posture Human perception of horizontal self(body)-motion in space was studied during various combinations of vestibular and leg-proprioceptive stimuli in the dark. During sinusoidal rotations of the trunk relative to the stationary feet (functionally synergistic combination) the perception was almost veridical over the frequency range tested (0.025~0.4 Hz). This finding suggested a dominance of the proprioceptive over the vestibular input, since the quantitative aspects of the perception (gain, phase, and detection threshold): (a) closely resembled those of the proprioceptive foot-to-trunk perception, and (b) clearly differed from those of the vestibular self-motion perception. However, when using other combinations, the self-motion perception changed in a monotonous way as a function of the two inputs, indicating that the two inputs do interact in a linear way. In a model of these findings the interaction occurs in two stages: (1) summation of a vestibular trunk-in-space signal and a (dynamically matched) proprioceptive foot-to-trunk signal yields an internal representation of foot support motion in space; (2) superposition of the latter by an almost ideal proprioceptive trunk-to-foot signal results in a representation of trunk-in-space motion (essentially proprioception-dependent and ideal when the feet are stationary).
A human observer in the dark, who is subjected to horizontal rotational accelerations of whole-body in space, registers head-in-space motion with the help of vestibular cues. Psychophysically, the evoked perception is non-veridical at low stimulus intensities, due to a high detection threshold [2]. In contrast, the perception is veridical during head rotation on the stationary trunk. This has been explained in terms of a two-stage vestibular-neck interaction: (1) the sum of high-threshold signals of trunk-to-head (neck) and head-in-space (vestibular) yields an internal notion of trunk-in-space rotation (zero during head rotation on stationary trunk); (2) a low-threshold nuchal head-on-trunk signal is superimposed on this, yielding a revised representation of head-in-space rotation [6]. The trunk, however, may be moved relative to the feet. In fact, it has been repeatedly suggested that the foot support may serve as a spatial reference. At the postural level, for instance, it has been suggested that equilibrium control centers use proprioceptive information from mechanoreceptors along the body axis to register head posi-
Correspondence: T. Mergner, Neurologische Universitatsklinik, Hansastr. 9, D-7800 Freiburg i.Br., FRG.
tion in relation to the body support [5]. At the perceptual level, a 'stand point' coordinate system has been suggested to explain the changes in orientation and location of visual objects when these are memorized after trunk and head displacements [7]. These considerations led us to study psychophysically the interaction between leg-proprioceptive and vestibular inputs for human perception of self-motion in space. The experimental setup consisted of a turning chair on which subjects (Ss) were seated for horizontal trunk rotation in space (since the head always was in fixed alignment with the trunk, this represented the vestibular stimulus, VEST). Their feet rested on a sled in front of them, which was driven by a position-controlled motor. It allowed the passive movement of the feet relative to the trunk (proprioceptive stimulus, PROP) on a circumferential path about the vertical body axis (radius, 0.5 m). The rotations were sinusoidal at 0.025, 0.05, 0.1, 0.2, and 0.4 Hz, peak angular displacement was +8 ° (if not specified otherwise). The stimuli were applied in the dark with the Ss' ears plugged. Six Ss participated in the study, each being tested twice with every stimulus. Their task was to indicate perceived trunk rotation in space with the help of a pointer (fixed to the chair, pivotable about the vertical body axis). To this
162
end, they performed during perceived body rotation a compensatory pointer rotation trying to keep it stationary in space. Position readings of the pointer rotations were fed, together with those of the chair and the foot sled, into a laboratory computer and stored for off-line analysis. As detailed in a previous study [6], indicated trunk rotation was related to its actual rotation and expressed in terms of gain and phase. After correction for the Ss' operator performance (manual ability of indication) and accounting for the fact that indication was opposite in direction to perceived self-motion, the data were taken as measures of the Ss' trunk-in-space perception. An analogous approach was used to evaluate the Ss' perception of foot-to-trunk rotation during PROP (using a 'foot' pointer). Furthermore, we evaluated their detection thresholds of trunk-in-space and foot-to-trunk rotation with the help of the pointer, in the same manner as described earlier [6]. In the following we shall present all measures in terms of median values and their 95% confidence intervals, since the individual data deviated from a normal distribution in some conditions. Ss" perception o f ' f o o t to trunk' during proprioceptive stimulation (foot rotation relative to the stationary trunk) is shown in Fig, 1A (upper panels). It had a gain of about unity across the frequency range tested and was roughly in-phase with foot position (phase ~ 0°). The threshold (lower panel), expressed in terms of both peak angular displacement and velocity, varied somewhat as a function of frequency, albeit in the opposite direction, over a range of 0.49-0.18 ° and 0.08-0.45°/s, respectively. Noticeably, these measures were almost identical to those previously reported for neck proprioception [6]. The results obtained for Ss' perception of 'trunk in space' during vestibular stimulation (trunk rotation together with the feet) essentially reproduced our previous findings [6]. The gain was only about 0.75 at 0.2 Hz and attenuated with decreasing frequency, reaching threshold at 0.025 Hz (Fig. I B). The phase showed a slight lead (re trunk position) at 0.2 Hz and advanced slightly with decreasing frequency. The threshold could be considered essentially a 'velocity threshold' on the order of 1-2°/s. Note that this threshold is higher by a factor of about 5 as compared to the aforementioned proprioceptive one. With the 'trunk in space' task the proprioceptive
stimulation elicited an illusion as if the trunk was being rotated in space. As shown in the 'gain' curve of Fig. 1C (gain re PROP) the illusion was weak at 0.1 Hz and increased with decreasing frequency. Its phase was approximately opposite to that of foot position (i.e. -180°). The illusion was missing at 0.2 and 0.4 Hz with the _+8° stimulus. However, it occurred at all frequencies during the threshold evaluation in ->50% of the trials. The threshold curves were essentially identical to those of the 'foot to trunk' perception (Fig. IA). When using a functionally synergistic combination of vestibular and proprioceptive stimuli (trunk rotation re stationary feet) Ss' perception of 'trunk in space' was approximately veridical (Fig. 1D). The threshold clearly was a purely proprioceptive one. The similarity of the gain, phase, and threshold curves with those of leg proprioception (cf. Fig. IA) suggested a dominance of the proprioceptive over the vestibular input during combination of the two inputs. However, as will be shown below, this notion does not hold. In a final experimental series, Ss were presented in random succession with stimuli of 5 different VESTPROP combinations (VEST=8°=constant, P R O P = - 16, -12, -8, -4, and 0 °, with negative sign of PROP indicating a phase difference between PROP and VEST of 180 ° ; VEST=8 °, PROP--0 ° and VEST=8 °, P R O P = - 8 ° represent repetitions of the stimuli used in Figs. I B and 1D, respectively). As shown in Fig 1E the gain (and phase, not shown) varied in a continuous way across the combinations, compatible with the notion of a linear summation of the two inputs. Our interpretation of the data is presented in the form of a model (Fig. 2). It is similar to that previously used to describe vestibular-neck interaction for human perception of head motion in space [6]. Vestibular-proprioceptive interaction in the model occurs in two stages: (i) figuratively speaking, the vestibular signal is channeled down to the feet to create an internal representation of loot-in-space rotation (high-threshold), to this end it is summed linearly with a proprioceptive signal, which is given the same response characteristics as the vestibular signal, (ii) superposition of a proprioceptive trunk-onfoot signal (low-threshold) on the loot-in-space representation: this yields the perception of 'trunk in space'. Note that the low-threshold proprioceptive signal alone
Fig. 1. A-D: perception of'foot to trunk' rotation during proprioceptive stimulation (A: PROP) and of 'trunk in space' during vestibular stimulation (B; VEST), PROP (C), and synergistic VEST-PROP combination (D). Gain and phase (upper panels) and detection threshold (lower panel) are given as a function of stimulus frequency (peak displacement, 8°). Pictographs show schematically the stimulus conditions. Percentages in C indicate frequency of occurrence of trunk motion illusion. Horizontal dashed lines give responses of an "ideal measuring device'. E: perception of 'trunk m space' as a function of both variations of VEST-PROP combination and stimulus frequency. Dotted plane shows simulated data from the model in Fig. 2.
163
" F ~ 0
B"TRUNKinSPACE"
. . . . . . .
10
10
0.5
0.5
z
3
0.5
~|
r~
0.025 0¢05
_o_
O~1
OT2
"TRUNKinSPACE"
C
0
0"4
qm
ql
*~
0.025 0"05
OT1
012
0.4
0,025
•-
o;5
o~1
0'2
04
w ~-
- -~-
---~.
- - -~
...-.~
-180
G.
c ~ PEAKDISPLACEMENT[o] e . . - o PEAK VELOCITY [°/s]
1.0
lO
5.0 "
% 67
..1 0 T
2.5
~ 0.5
~O
O .
p
0.025
0.05
,
,qln
0.1
0.2
0.4
4m
0.025
•~
v
0.05
0:1
012
0
m.
0.025
1
u
0.05
E "TRUNK in SPACE"
•
•
. D
0.1
0.2
0.4
FREQUENCY[Hz]
Z_
o1 1.0 a
~0.5 n"r I-
jy"
O.
a=
0.025 0.65
o:1
i
0.2
.D
0.4
0.1
mw
0.025
0.4
PROP~
z 0.5 3
0.()5
50
71
_
0 ,
D "TRUNK in SPACE"
~45
b
-
,
83
0.5
I--
m,
75
O 0.025
•
m
0.2
v III
0.4
164 VESTIBULAR SYS.
+
HT:O + CHAIa: TRUNKTs _1 in ¢ ~ SPACE
FOOT in o SPACE
FS
I ,
-1
PROPR. SYS.
+'t's' n
r
TS
I L~-JL?J t + ~ . J ", + ~ fs
Fig. 2. Model of vestibular-proprioceptive interaction for the perception of'trunk in space' (7q'S). HS, TS, HT, FT, and FS: head-in-space, trunk-in-space, head-to-trunk, foot-to-trunk, and foot-in-space displacement, respectively; hs, t's, h t , r t and rs: internal representations of these displacements as velocity signals. The interaction is considered to occur in two stages: (i) internal reconstruction of foot motion in space, fs, resulting from the sum of t s and ( t (ft is high-pass filtered, r=5s, to match the dynamics of hs) and (ii) summation of rs and - f t (non-filtered; low threshold, 0.15°/s), resulting in a revised version of t s (ts'). t s ' undergoes a gain attenuation (G2-0.75; assumed to occur with repetitive stimulation) and a final integration before it yields 7q'S. As detailed earlier [7], the time constant of rs is prolonged to r=20s by a low frequency recovery mechanism (a), and stability is maintained by a velocity threshold of 1°/s (b). ~ and f, differentiation and integration, respectively.
determines the 'trunk in space' perception when the feet are stationary (compare experimental and simulated data with VEST=8 °, P R O P = - 8 ° in Fig. 1E). On the other hand, the proprioceptively induced illusion of trunk-in-space motion (Fig. 1C) results from the fact that the two proprioceptive signals in the model are opposite in sign, but do not cancel each other completely, mainly because they differ with respect to their thresholds. Our findings on leg proprioception are analogous in essence to those we previously reported on the role of neck proprioception for the perception of head motion in space [6]. Thus, it appears that, perceptually, the vestibular information is linked to the foot support representation with the help of proprioceptive inputs, wherever these arise along the body axis. This notion has two implications. First, the perception can be considered as representing a stack of coordinate systems, built bottomup with the feet related to space coordinates (vestibularly derived), the trunk related to the feet, and the head to the trunk. Second, the perception uses the vestibular input to control for stationarity of external references, and, if true, relies on these instead of the vestibularly derived space reference, thus avoiding the disadvantage of the high vestibular threshold.
This concept may possibly apply also to equilibrium control. For instance, the direction of vestibularly induced postural sway is determined by the horizontal excursion between head and foot, independently of whether the excursion is produced at the level of the neck or the legs [5]. Convergence and interaction of vestibular and peripheral nerve inputs are not only found in several brain structures, but also at spinal levels, from where the result is conveyed bottom-up [1]. A bottom-up activation pattern from the legs to paraspinal muscles has been observed in postural responses to small translation of the foot support [3]. Finally, we have to point out that there is a mode of 'trunk in space' perception that does not rely on the vestibular-proprioceptive interaction as outlined above. With large foot-to-trunk excursions (>8°), Ss appeared to rely predominantly on the vestibular cues, as if ignoring the foot support as a reference. The latter also applied if Ss focused attention on their feet rather than on their trunks. In our model (Fig. 2) this could be simulated by a gain attenuation of the proprioceptive signals. We thus conceive of two perceptual modes, one relying predominantly on vestibular cues and the other on vestibularproprioceptive interaction (with apparent proprioceptive dominance when the feet are perceptually stationary). Possibly, this notion has an analogy in postural control with a predominantly vestibular-governed 'hip strategy' and proprioceptive-governed 'ankle strategy' [4]. Supported by DFG, SFB 325 and CSR-113/27. 1 Coulter, J.D., Mergner, T. and Pompeiano, O., Effects of static tilt on cervical spinoreticular tract neurons, J. Neurophysiol., 39 (1976) 45 62. 2 Guedry, F.E., Psychophysics of vestibular sensation. In H.H. Kornh uber (Ed.), Handbook of Sensory Physiology, Vol. V]/2, Vestibular System, Springer, Berlin, 1974, pp. 1-154. 3 Horak, F.B. and Nashner, L.M., Central programming of postural movements: adaptation to altered support-surface configurations, J. Neurophysiol., 55 (1986) 1369 1381. 4 Horak, F.B., Nashner, L.M. and Diener, H.C., Postural strategies associated with somatosensory and vestibular loss, Exp. Brain Res., 82(1990) 167 177. 5 Lund, S. and Broberg, C., Effects of different head positions on postural sway in man induced by a reproducible vestibular error signal, Acta Physiol. Scan&, 117 (1983) 307 309. 6 Mergner, T., Siebold, C., Schweigart, G. and Becker, W., Human perception of horizontal trunk and head rotation in space during vestibular and neck stimulation, Exp. Brain Res., 85 (1991) 389 404. 7 MUller, G.E., Zur Analyse der Ged~ichtnist~tigkeit und des Vorstellungsverlaufes. Vol. 2, Johann Ambrosius Barth, Leipzig, 1917.
161
NSL 08549
Interaction of vestibular and proprioceptive inputs for human self-motion perception F. Hlava6ka a, T. Mergner b and G. Schweigart b alnstitute of Normal and Pathological Physiology, Slovak Academy of Science, Bratislava (Czechoslovakia) and bNeurological University Clinic, Freiburg ( FRG ) (Received 6 December 1991; Revised version received 20 January 1992; Accepted 22 January 1992)
Key words: Human; Perception; Self-motion; Vestibular stimulation; Proprioceptive stimulation; Posture Human perception of horizontal self(body)-motion in space was studied during various combinations of vestibular and leg-proprioceptive stimuli in the dark. During sinusoidal rotations of the trunk relative to the stationary feet (functionally synergistic combination) the perception was almost veridical over the frequency range tested (0.025~0.4 Hz). This finding suggested a dominance of the proprioceptive over the vestibular input, since the quantitative aspects of the perception (gain, phase, and detection threshold): (a) closely resembled those of the proprioceptive foot-to-trunk perception, and (b) clearly differed from those of the vestibular self-motion perception. However, when using other combinations, the self-motion perception changed in a monotonous way as a function of the two inputs, indicating that the two inputs do interact in a linear way. In a model of these findings the interaction occurs in two stages: (1) summation of a vestibular trunk-in-space signal and a (dynamically matched) proprioceptive foot-to-trunk signal yields an internal representation of foot support motion in space; (2) superposition of the latter by an almost ideal proprioceptive trunk-to-foot signal results in a representation of trunk-in-space motion (essentially proprioception-dependent and ideal when the feet are stationary).
A human observer in the dark, who is subjected to horizontal rotational accelerations of whole-body in space, registers head-in-space motion with the help of vestibular cues. Psychophysically, the evoked perception is non-veridical at low stimulus intensities, due to a high detection threshold [2]. In contrast, the perception is veridical during head rotation on the stationary trunk. This has been explained in terms of a two-stage vestibular-neck interaction: (1) the sum of high-threshold signals of trunk-to-head (neck) and head-in-space (vestibular) yields an internal notion of trunk-in-space rotation (zero during head rotation on stationary trunk); (2) a low-threshold nuchal head-on-trunk signal is superimposed on this, yielding a revised representation of head-in-space rotation [6]. The trunk, however, may be moved relative to the feet. In fact, it has been repeatedly suggested that the foot support may serve as a spatial reference. At the postural level, for instance, it has been suggested that equilibrium control centers use proprioceptive information from mechanoreceptors along the body axis to register head posi-
Correspondence: T. Mergner, Neurologische Universitatsklinik, Hansastr. 9, D-7800 Freiburg i.Br., FRG.
tion in relation to the body support [5]. At the perceptual level, a 'stand point' coordinate system has been suggested to explain the changes in orientation and location of visual objects when these are memorized after trunk and head displacements [7]. These considerations led us to study psychophysically the interaction between leg-proprioceptive and vestibular inputs for human perception of self-motion in space. The experimental setup consisted of a turning chair on which subjects (Ss) were seated for horizontal trunk rotation in space (since the head always was in fixed alignment with the trunk, this represented the vestibular stimulus, VEST). Their feet rested on a sled in front of them, which was driven by a position-controlled motor. It allowed the passive movement of the feet relative to the trunk (proprioceptive stimulus, PROP) on a circumferential path about the vertical body axis (radius, 0.5 m). The rotations were sinusoidal at 0.025, 0.05, 0.1, 0.2, and 0.4 Hz, peak angular displacement was +8 ° (if not specified otherwise). The stimuli were applied in the dark with the Ss' ears plugged. Six Ss participated in the study, each being tested twice with every stimulus. Their task was to indicate perceived trunk rotation in space with the help of a pointer (fixed to the chair, pivotable about the vertical body axis). To this
162
end, they performed during perceived body rotation a compensatory pointer rotation trying to keep it stationary in space. Position readings of the pointer rotations were fed, together with those of the chair and the foot sled, into a laboratory computer and stored for off-line analysis. As detailed in a previous study [6], indicated trunk rotation was related to its actual rotation and expressed in terms of gain and phase. After correction for the Ss' operator performance (manual ability of indication) and accounting for the fact that indication was opposite in direction to perceived self-motion, the data were taken as measures of the Ss' trunk-in-space perception. An analogous approach was used to evaluate the Ss' perception of foot-to-trunk rotation during PROP (using a 'foot' pointer). Furthermore, we evaluated their detection thresholds of trunk-in-space and foot-to-trunk rotation with the help of the pointer, in the same manner as described earlier [6]. In the following we shall present all measures in terms of median values and their 95% confidence intervals, since the individual data deviated from a normal distribution in some conditions. Ss" perception o f ' f o o t to trunk' during proprioceptive stimulation (foot rotation relative to the stationary trunk) is shown in Fig, 1A (upper panels). It had a gain of about unity across the frequency range tested and was roughly in-phase with foot position (phase ~ 0°). The threshold (lower panel), expressed in terms of both peak angular displacement and velocity, varied somewhat as a function of frequency, albeit in the opposite direction, over a range of 0.49-0.18 ° and 0.08-0.45°/s, respectively. Noticeably, these measures were almost identical to those previously reported for neck proprioception [6]. The results obtained for Ss' perception of 'trunk in space' during vestibular stimulation (trunk rotation together with the feet) essentially reproduced our previous findings [6]. The gain was only about 0.75 at 0.2 Hz and attenuated with decreasing frequency, reaching threshold at 0.025 Hz (Fig. I B). The phase showed a slight lead (re trunk position) at 0.2 Hz and advanced slightly with decreasing frequency. The threshold could be considered essentially a 'velocity threshold' on the order of 1-2°/s. Note that this threshold is higher by a factor of about 5 as compared to the aforementioned proprioceptive one. With the 'trunk in space' task the proprioceptive
stimulation elicited an illusion as if the trunk was being rotated in space. As shown in the 'gain' curve of Fig. 1C (gain re PROP) the illusion was weak at 0.1 Hz and increased with decreasing frequency. Its phase was approximately opposite to that of foot position (i.e. -180°). The illusion was missing at 0.2 and 0.4 Hz with the _+8° stimulus. However, it occurred at all frequencies during the threshold evaluation in ->50% of the trials. The threshold curves were essentially identical to those of the 'foot to trunk' perception (Fig. IA). When using a functionally synergistic combination of vestibular and proprioceptive stimuli (trunk rotation re stationary feet) Ss' perception of 'trunk in space' was approximately veridical (Fig. 1D). The threshold clearly was a purely proprioceptive one. The similarity of the gain, phase, and threshold curves with those of leg proprioception (cf. Fig. IA) suggested a dominance of the proprioceptive over the vestibular input during combination of the two inputs. However, as will be shown below, this notion does not hold. In a final experimental series, Ss were presented in random succession with stimuli of 5 different VESTPROP combinations (VEST=8°=constant, P R O P = - 16, -12, -8, -4, and 0 °, with negative sign of PROP indicating a phase difference between PROP and VEST of 180 ° ; VEST=8 °, PROP--0 ° and VEST=8 °, P R O P = - 8 ° represent repetitions of the stimuli used in Figs. I B and 1D, respectively). As shown in Fig 1E the gain (and phase, not shown) varied in a continuous way across the combinations, compatible with the notion of a linear summation of the two inputs. Our interpretation of the data is presented in the form of a model (Fig. 2). It is similar to that previously used to describe vestibular-neck interaction for human perception of head motion in space [6]. Vestibular-proprioceptive interaction in the model occurs in two stages: (i) figuratively speaking, the vestibular signal is channeled down to the feet to create an internal representation of loot-in-space rotation (high-threshold), to this end it is summed linearly with a proprioceptive signal, which is given the same response characteristics as the vestibular signal, (ii) superposition of a proprioceptive trunk-onfoot signal (low-threshold) on the loot-in-space representation: this yields the perception of 'trunk in space'. Note that the low-threshold proprioceptive signal alone
Fig. 1. A-D: perception of'foot to trunk' rotation during proprioceptive stimulation (A: PROP) and of 'trunk in space' during vestibular stimulation (B; VEST), PROP (C), and synergistic VEST-PROP combination (D). Gain and phase (upper panels) and detection threshold (lower panel) are given as a function of stimulus frequency (peak displacement, 8°). Pictographs show schematically the stimulus conditions. Percentages in C indicate frequency of occurrence of trunk motion illusion. Horizontal dashed lines give responses of an "ideal measuring device'. E: perception of 'trunk m space' as a function of both variations of VEST-PROP combination and stimulus frequency. Dotted plane shows simulated data from the model in Fig. 2.
163
" F ~ 0
B"TRUNKinSPACE"
. . . . . . .
10
10
0.5
0.5
z
3
0.5
~|
r~
0.025 0¢05
_o_
O~1
OT2
"TRUNKinSPACE"
C
0
0"4
qm
ql
*~
0.025 0"05
OT1
012
0.4
0,025
•-
o;5
o~1
0'2
04
w ~-
- -~-
---~.
- - -~
...-.~
-180
G.
c ~ PEAKDISPLACEMENT[o] e . . - o PEAK VELOCITY [°/s]
1.0
lO
5.0 "
% 67
..1 0 T
2.5
~ 0.5
~O
O .
p
0.025
0.05
,
,qln
0.1
0.2
0.4
4m
0.025
•~
v
0.05
0:1
012
0
m.
0.025
1
u
0.05
E "TRUNK in SPACE"
•
•
. D
0.1
0.2
0.4
FREQUENCY[Hz]
Z_
o1 1.0 a
~0.5 n"r I-
jy"
O.
a=
0.025 0.65
o:1
i
0.2
.D
0.4
0.1
mw
0.025
0.4
PROP~
z 0.5 3
0.()5
50
71
_
0 ,
D "TRUNK in SPACE"
~45
b
-
,
83
0.5
I--
m,
75
O 0.025
•
m
0.2
v III
0.4
164 VESTIBULAR SYS.
+
HT:O + CHAIa: TRUNKTs _1 in ¢ ~ SPACE
FOOT in o SPACE
FS
I ,
-1
PROPR. SYS.
+'t's' n
r
TS
I L~-JL?J t + ~ . J ", + ~ fs
Fig. 2. Model of vestibular-proprioceptive interaction for the perception of'trunk in space' (7q'S). HS, TS, HT, FT, and FS: head-in-space, trunk-in-space, head-to-trunk, foot-to-trunk, and foot-in-space displacement, respectively; hs, t's, h t , r t and rs: internal representations of these displacements as velocity signals. The interaction is considered to occur in two stages: (i) internal reconstruction of foot motion in space, fs, resulting from the sum of t s and ( t (ft is high-pass filtered, r=5s, to match the dynamics of hs) and (ii) summation of rs and - f t (non-filtered; low threshold, 0.15°/s), resulting in a revised version of t s (ts'). t s ' undergoes a gain attenuation (G2-0.75; assumed to occur with repetitive stimulation) and a final integration before it yields 7q'S. As detailed earlier [7], the time constant of rs is prolonged to r=20s by a low frequency recovery mechanism (a), and stability is maintained by a velocity threshold of 1°/s (b). ~ and f, differentiation and integration, respectively.
determines the 'trunk in space' perception when the feet are stationary (compare experimental and simulated data with VEST=8 °, P R O P = - 8 ° in Fig. 1E). On the other hand, the proprioceptively induced illusion of trunk-in-space motion (Fig. 1C) results from the fact that the two proprioceptive signals in the model are opposite in sign, but do not cancel each other completely, mainly because they differ with respect to their thresholds. Our findings on leg proprioception are analogous in essence to those we previously reported on the role of neck proprioception for the perception of head motion in space [6]. Thus, it appears that, perceptually, the vestibular information is linked to the foot support representation with the help of proprioceptive inputs, wherever these arise along the body axis. This notion has two implications. First, the perception can be considered as representing a stack of coordinate systems, built bottomup with the feet related to space coordinates (vestibularly derived), the trunk related to the feet, and the head to the trunk. Second, the perception uses the vestibular input to control for stationarity of external references, and, if true, relies on these instead of the vestibularly derived space reference, thus avoiding the disadvantage of the high vestibular threshold.
This concept may possibly apply also to equilibrium control. For instance, the direction of vestibularly induced postural sway is determined by the horizontal excursion between head and foot, independently of whether the excursion is produced at the level of the neck or the legs [5]. Convergence and interaction of vestibular and peripheral nerve inputs are not only found in several brain structures, but also at spinal levels, from where the result is conveyed bottom-up [1]. A bottom-up activation pattern from the legs to paraspinal muscles has been observed in postural responses to small translation of the foot support [3]. Finally, we have to point out that there is a mode of 'trunk in space' perception that does not rely on the vestibular-proprioceptive interaction as outlined above. With large foot-to-trunk excursions (>8°), Ss appeared to rely predominantly on the vestibular cues, as if ignoring the foot support as a reference. The latter also applied if Ss focused attention on their feet rather than on their trunks. In our model (Fig. 2) this could be simulated by a gain attenuation of the proprioceptive signals. We thus conceive of two perceptual modes, one relying predominantly on vestibular cues and the other on vestibularproprioceptive interaction (with apparent proprioceptive dominance when the feet are perceptually stationary). Possibly, this notion has an analogy in postural control with a predominantly vestibular-governed 'hip strategy' and proprioceptive-governed 'ankle strategy' [4]. Supported by DFG, SFB 325 and CSR-113/27. 1 Coulter, J.D., Mergner, T. and Pompeiano, O., Effects of static tilt on cervical spinoreticular tract neurons, J. Neurophysiol., 39 (1976) 45 62. 2 Guedry, F.E., Psychophysics of vestibular sensation. In H.H. Kornh uber (Ed.), Handbook of Sensory Physiology, Vol. V]/2, Vestibular System, Springer, Berlin, 1974, pp. 1-154. 3 Horak, F.B. and Nashner, L.M., Central programming of postural movements: adaptation to altered support-surface configurations, J. Neurophysiol., 55 (1986) 1369 1381. 4 Horak, F.B., Nashner, L.M. and Diener, H.C., Postural strategies associated with somatosensory and vestibular loss, Exp. Brain Res., 82(1990) 167 177. 5 Lund, S. and Broberg, C., Effects of different head positions on postural sway in man induced by a reproducible vestibular error signal, Acta Physiol. Scan&, 117 (1983) 307 309. 6 Mergner, T., Siebold, C., Schweigart, G. and Becker, W., Human perception of horizontal trunk and head rotation in space during vestibular and neck stimulation, Exp. Brain Res., 85 (1991) 389 404. 7 MUller, G.E., Zur Analyse der Ged~ichtnist~tigkeit und des Vorstellungsverlaufes. Vol. 2, Johann Ambrosius Barth, Leipzig, 1917.
