Motion processing

Acceleration is detected by comparing initial and final velocities

Brian Timney, Samantha Kearney and Benjamin Asa

There are many studies of constant velocity motion, but few on the perception of acceleration in the fronto-parallel plane. Logically, acceleration may be detected in two ways: first, it may be processed directly by "acceleration detectors" in the cortex tuned to specific acceleration rates; second, there may be an indirect process whereby acceleration is “inferred” by comparing initial and final velocities. We tested these alternatives by measuring acceleration thresholds while varying absolute acceleration rates within a single presentation. We designed a three-component stimulus in which the initial and final rates were identical, and a middle component that could have a negative, positive, or zero acceleration. We measured acceleration thresholds by varying the initial/final rates using a constant stimuli procedure. If acceleration is detected directly, the initial/final rate should determine thresholds, regardless of the middle rate; if it is detected indirectly, the difference between initial and final velocities, and hence the average acceleration rate across the whole presentation, should be the determining factor. While initial/final rate thresholds changed markedly as the middle rate was varied, these differences were eliminated when the data were replotted as a function of average acceleration. This is strong evidence for an indirect mechanism of acceleration detection.

Short term memory for visual coherent motion direction revealed by visual masking

Andrea Pavan, Dominik Langgartner and Mark W. Greenlee

Psychophysical research has shown that different attributes of visual stimuli, such as spatial frequency, orientation and speed can be stored in visual short term memory (VSTM) with a high degree of accuracy. Visual memory masking paradigms have been extensively employed to determine which attributes are important in the storage of the information in VSTM. We examined the properties of the storage mechanism for coherent motion. In particular, we presented coherent motion in two distinct temporal intervals, i.e., memory and test stimuli (duration: 0.2 s), and introducing random-motion or coherent-motion mask in a 3-s delay. The results (N = 8) showed that mask mainly interferes with performance when displayed 0.2 s after the offset of the sample and when it was directional rather than random. When using directional masking, we observed a drop in accuracy of ~11% with respect to the no-masking condition. These results suggest that the memory representation of coherent motion is selective for direction, being compromised by intervening directional stimuli presented immediately after the encoding phase [Pasternak & Zaksas, 2003, J Neurophysiol, 90, 2757–2762], and that the neural mechanisms involved in the processing of coherent motion may also be involved in its storage.

Perceived motion of moving barber pole arrays is determined by a streaming process

George Sperling, Peng Sun and Charles Chubb

We introduce a class of stimuli that demonstrate a new motion perception process. The stimuli are diagonal sinusoidal carrier gratings with bars drifting up to the left, windowed by a raised, vertical, drifting sinusoid. In slow motion, viewed foveally, these stimuli are perceived veridically as arrays of barber poles with blurred edges, the whole array moving laterally (perpendicularly to the orientation of the barber poles) while the stripes inside the poles move up. However, when viewed peripherally, the perceived motion is unambiguously vertical (up or down depending on the direction of motion of stripes within the barber poles) over a wide range of stripe speeds and array speeds. This is surprising because: (1) this stimulus actually moves rigidly in a diagonal direction (feature-tracking direction); (2) the dominant components and vector average of the Fourier components of this stimulus are also diagonal (3) combinations of (1) and (2) are not vertical; (4) second-order motion is shown to be irrelevant. Conclusion: The direction of perceived motion is determined by the orientation of the stream-beds (barber poles) within which pattern motion flows. This streaming process is critically sensitive to stream-bed orientation but remarkably insensitive to the lateral motion of the stream-bed itself.

Integration of visual-vestibular information for self-motion perception: Role of posterior insular cortex

Mark W. Greenlee, Sebastian M. Frank, Oliver Baumann and Jason Mattingley

Self-motion perception relies on cues from visual and vestibular senses. Extracellular recordings in the parieto-insular cortex (PIVC; Chen et al. 2010, J Neurosc. 30, 3022) and the posterior sylvian fissure (VPS; Chen et al 2011, J Neurosc. 31, 11617) in primates indicate different roles of these regions in the processing of self-motion cues. Extending our first report (Greenlee et al., 2011, Perception 40 ECVP Abstract Suppl., p. 26), we analysed the selectivity of these regions in human cortex for visual and vestibular stimulation. Nine participants viewed a large (30 deg) display containing random-dot motion (white dots on dark background). A subset of the dots (10%) moved either to the right or left. During the presentation of coherent motion bithermic caloric irrigation of both ear canals induced a near-threshold sensation of head rotation (yaw) either to the left or right. In a 4-AFC paradigm, participants responded, whether they sensed self motion (in-phase or out-of-phase with visual motion) or object motion (left or right directions). In a control condition, caloric stimulation was applied in the dark (vestibular stimulation only). Here we present the functional analysis of selected ROIs (MST, motion-sensitive STS, PIVC and PIC) that indicated significantly enhanced activation during self-motion in visual-vestibular and vestibular-only conditions. Area MST and STS respond to visual/vestibular stimulation and more to their combination. Area PIVC responds primarily to vestibular, while PIC responds well to visual, vestibular and combined visual-vestibular stimulation. These findings suggest that PIC may be the human homologue of primate VPS, and it supports the integration of visual-vestibular information in self-motion perception. (Support: DAAD—Go8 Exchange Programme)

Disparate persistence of illusory depth and illusory motion in structure-from-motion displays

Jochen Braun and Alexander Pastukhov

Ambiguous illusory percepts tend to persist even when the display is interrupted and this persistence can be used to probe the specificity of memory representations. We used a structure-from-motion display and dissociated the illusory depth of the interpolated shape from its illusory rotation via the use of a rotationally asymmetric shape (an illusory band) and reversals of physical planar motion. Latter ensured that the prior percept could persist either in terms of illusory rotation or illusory depth of the shape, but not both [Pastukhov, Vonau and Braun, 2012, Journal of Vision 12(1):17]. Observers viewed an ambiguously rotating band and reported on the perceived rotation both before and after interruptions of variable duration. After interruptions of 0.5 seconds or more, the previous illusory motion resumed consistently, demonstrating the superior persistence of the associated memory traces. However, after shorter interruptions, perceptual dominance was determined by the previous illusory depth. This short-lived memory trace is static (showing no sign of dynamical updating) and is specific to the interpolated illusory shape. It is also erased by masking (resembling visual sensnory memory). We conclude that disparate visual persistence contributes further evidence for distinct representations of illusory rotation and illusory depth in structure-from-motion displays.

Motion direction integration following the onset of multistable stimuli (II): stability properties explain dynamic shifts in the dominant perceived direction

James Rankin, Andrew I. Meso, Guillaume S. Masson, Olivier Faugeras and Pierre Kornprobst

We use a mathematical model to investigate the early temporal dynamics of motion integration with a multistable input. A drifting grating stimulus is considered with an aperture configuration that supports horizontal (H), diagonal (D) or vertical (V) perceived directions for the same input. A shift in perceptual dominance from D to either H/V with increased presentation duration has been shown in short-presentation psychophysics experiments; see our companion abstract (I). We work with a neural fields, population-level representation of cortical activity that performs motion integration of the competing oblique (D) and cardinal (H/V) local direction signals [Rankin et al., 2011, INRIA Report 7822]. When a dynamical system diverges away from a weakly unstable state it can spend an extended transient period close to the unstable state before divergence [Strogatz, 2001, Nonlinear Dynamics and Chaos, Perseus, NY]. Of the competing states representing the alternative percepts, the initially dominant D is computed to be weakly unstable whilst the subsequently dominant H/V are computed to be stable. The property of weak stability explains the gradual shift in dominance from D to H/V. Moreover, as contrast increases, D becomes strongly unstable reducing the latency of this shift in perceptual dominance.