Vision plays an important role in balance control for standing and walking by providing information about movement relative to the environment via optical flow (Gibson, 1958). Quantifying the capabilities of the human visual system is challenging, as any particular property such as contrast sensitivity (Owsley, 2003), depth perception (Walk and Gibson, 1961; Brenner and Smeets, 2018), and motion detection (Borst and Egelhaaf, 1989) may affect different functional behaviors. The ability to detect self-motion from optical flow is expected to be most relevant for balance control, but visual motion detection thresholds are typically measured during sitting (Warren et al., 1989; Gilmore et al., 1992; Turano and Wong, 1992; Habak and Faubert, 2000; Freeman et al., 2006, 2008; Conlon et al., 2017), where self-motion is eliminated, and balance is not an issue. Our motivation for this study was to investigate whether: (1) visual motion thresholds can be reliably measured during standing and walking; and (2) to determine whether thresholds differ during balance tasks when self-motion is not constrained.

When studies investigate the relationship between visual processing and fall risk, they typically assess qualities of visual acuity such as contrast sensitivity (Lord and Fitzpatrick, 2001; Wood et al., 2011), depth perception (Felson et al., 1989; Lord and Dayhew, 2001), or size of the visual field (Ivers et al., 1998; Broman et al., 2004). Visual acuity is meaningful for maneuvering around an environment and avoiding falls caused by tripping or hitting obstacles (Broman et al., 2004) as vision provides information about object size, location, and where to place the swing leg foot into a safe space. Visual acuity relates to central vision, or focal vision, capable of high spatial resolution and particularly useful for pattern and object recognition (Larson and Loschky, 2009). While visual acuity mostly concerns central vision, visual motion perception is more related to peripheral vision (Monaco et al., 2007). Illusion of self-motion in response to visual motion, “vection,” has been shown to be primarily influenced by stimuli in the peripheral visual field (Brandt et al., 1973; Tarita-Nistor et al., 2008, 2014). Optic flow can produce illusions of self-motion, and thus disturb upright balance in both standing (Peterka, 2002; Kiemel et al., 2006; Jeka et al., 2010) and walking (Logan et al., 2010; McAndrew et al., 2010; Franz et al., 2015; Reimann et al., 2018). To our knowledge, there has been only one study that has directly compared measures of visual acuity to motion perception in their relationship to control of upright balance. Data collected during the Salisbury Eye Evaluation (SEE Project) (Freeman et al., 2008) found that in a model including visual acuity, contrast sensitivity, visual field, and motion detection threshold, the motion detection thresholds were associated with over three times higher odds of failing on a single leg balance stance task when adjusted for age, sex, and race compared to the other measures of vision. A review by Saftari and Kwon (2018) highlights the general finding that decreased visual acuity is associated with increased risk for falls and hip fractures. Despite these findings, they emphasize that visual motion perception as a contributor to fall risk has been a critical omission in the literature.

Here we tested the reliability of measuring a visual motion detection threshold for detecting optic flow during standing and walking, tasks typically performed for investigating upright balance control. To our knowledge, visual motion detection tests have only been performed while sitting in which the head is typically immobilized (Warren et al., 1989; Gilmore et al., 1992; Turano and Wong, 1992; Habak and Faubert, 2000; Freeman et al., 2006, 2008; Conlon et al., 2017). Here we measured visual motion detection thresholds during standing and walking, where body sway generates a natural background level of self-motion. A threshold measure characterizes the underlying mechanism (Kingdom and Prins, 2009) of a sensory system. In the case of a visual motion detection test, this threshold provides a measure of how sensitive the visual system is in detecting movement in the environment. To calculate a threshold value, perceptual responses are recorded after exposing participants to optic flow stimuli with varying directions and maximum amplitudes. In this study, we use a common adaptive psychophysical method, in which the amplitude of the stimulus is increased or decreased depending on the history of responses. Our hypotheses are that (1) visual motion detection thresholds are correlated between repeated measures in both standing and walking and (2) thresholds in walking are different than in standing.

Twenty-nine healthy participants (14 female, 39 ± 15 years old) between the ages of 23 and 67 were recruited for this experiment. Subjects provided informed verbal and written consent to participate. Subjects did not have a history of neurological disorders or visual diagnoses, and no history of surgical procedures involving the legs, spine, or head within 6 months of the protocol. Participants presented with normal or corrected to normal vision (glasses/contacts). The experiment was approved by the University of Delaware Institutional Review Board.

All subjects completed the experiment of both walking and standing conditions. On the individual level, 24 subjects had higher thresholds during walking versus standing. The average walking speed for participants during the walking condition was 1.09 m/s (SD = 0.20 m/s).

Our study investigated the reliability of measuring a visual motion detection threshold during walking and standing and compared these measures between those two tasks. We used virtual reality to display a visual scene that rotated in a clockwise or counter-clockwise motion at different amplitudes and asked participants to report the direction of rotation. We found that a visual motion threshold can be reliably obtained during tasks involving body sway and movement, namely standing and walking. We also found evidence that visual motion detection thresholds are higher when walking versus standing, potentially influencing the way visual information is processed between the two tasks.

Visual motion detection thresholds can be reliably obtained during walking and standing. Strong correlations indicate good agreement between the two measures taken at different time points. Bland–Altman plots in Figure 4 support this finding by indicating small average differences between the two measurements for both walking (0.16°) and standing (0.01°). Traditionally, visual motion detection tests are performed while sitting, with the head fixated to avoid any type of head movement (Warren et al., 1989; Gilmore et al., 1992; Turano and Wong, 1992; Wood and Bullimore, 1995; Tran et al., 1998; Habak and Faubert, 2000; Snowden and Kavanagh, 2006; Freeman et al., 2006, 2008; Conlon et al., 2017). These tests have explored different aspects of motion detection and it is unclear which are most relevant for balance. For example, minimal displacement thresholds have been quantified using the translation of a random dot pattern (Habak and Faubert, 2000; Snowden and Kavanagh, 2006; Conlon et al., 2017), motion coherence thresholds have been quantified by translating select percentages of those dots (Gilmore et al., 1992; Tran et al., 1998; Snowden and Kavanagh, 2006), speed discrimination thresholds have been quantified by varying the object motion’s speed (Snowden and Kavanagh, 2006), and heading direction thresholds have been quantified by varying optic flow patterns relevant to a vertical object (Warren and Hannon, 1988; Warren et al., 1989). These studies all implement psychophysical testing, however, there is no standardized approach, using different methods such as the BEST PEST (Gilmore et al., 1992; Habak and Faubert, 2000), method of constant stimuli (Turano and Wong, 1992; Tran et al., 1998), or adaptive staircases (Wood and Bullimore, 1995; Snowden and Kavanagh, 2006; Freeman et al., 2006, 2008). Furthermore, these studies have manipulated and controlled for various characteristics of the stimuli such as luminance and contrast. Again, it is unclear which of these aspects of motion discrimination is most relevant to balance, thus motivating our study to see if motion discrimination is viable to explore during upright balance tasks such as standing and walking with the inherent added head movement. Immobilizing the head eliminates retinal slip caused by natural head sway during standing and walking. This may explain why values of motion discrimination are considerably smaller while sitting (Turano and Wong, 1992; Snowden and Kavanagh, 2006; Freeman et al., 2006, 2008; Conlon et al., 2017), obtaining values ranging from 0.009° to 0.121°. Heading direction thresholds obtained by Warren and Hannon (1988), and Warren et al. (1989) may be more relevant to upright balance control since navigating through our environment via optic flow is important for locomotion. Thresholds calculated via heading direction are similar to what is seen here, ranging from 1.1° to 1.9°. Although it must be noted that motion thresholds here cannot be directly compared to past literature due to methodological differences such as psychophysical protocols, and stimulus conditions such as luminance, contrast, and type of motion. Here, we used a rotation of the virtual scene to measure motion detection. This type of stimulus was selected based on previous work investigating the role of vision and balance during walking in which visual perturbations have been implemented around the anterior/posterior axis to simulate the sensation of a fall and investigate medial/lateral balance control (McAndrew et al., 2010; Franz et al., 2015; Reimann et al., 2018).

The visual motion stimulus presented here is in terms of maximum angular amplitude per 1 s, resulting in changes of both displacement and velocity of the stimulus during the adaptive staircase protocol to quantify a visual motion threshold. Visual motion processing has been generally viewed as having two distinct processes: first and second order processing. First order processing is defined by differences in luminance while second order processing is defined by differences in contrast, texture, or depth (Seiffert and Cavanagh, 1998; Habak and Faubert, 2000; Baker and Mareschal, 2001). Seiffert and Cavanagh (1998) have shown that second order processing is position based and support findings from Nakayama and Tyler (1981) that first order processing is velocity based. Since both speed and amplitude were manipulated in this experiment, we expect that a combination of first order and second order visual processing were used to detect the movement of the stimuli. It is still undetermined as to which parameter is more or less useful for controlling upright balance. To note, it has been suggested that the loss of accurate velocity estimations from a sensory modality based on modeling work from Kiemel et al. (2002) are more problematic to upright postural control than position estimates (Jeka et al., 2004). Snowden and Kavanagh (2006) provide evidence that speed discrimination is hindered in older adults compared to healthy young. The loss of velocity dependent information in visual motion processing could potentially be an underlying mechanism to hindered upright balance control. Since the stimulus presented here contains both a position and speed component, the underlying neural mechanism cannot be attributed to one or the other.

Our motivation was to perform the psychometric tests in an ergonomic manner that would quantify visual motion thresholds during natural movements of the head and body while maintaining upright balance, as opposed to a restrained position. Since participants were not restrained in any way, potential cues from the vestibular and proprioceptive system from the natural motion may have influenced the participants’ threshold results. Although small, a visual stimulus can provoke the illusion of a fall and cause a balance response that may cause added movement of the head and/or ankles, cueing the vestibular or proprioceptive system (Reimann et al., 2018). Some subjects reported perception of self-motion when the visual stimulus became small, rather than movement of the visual stimulus. Such cues from other sensory systems may have influenced responses that could not be considered purely visual. Head movements in standing are relatively small, but walking produces considerable head movement and might influence a person’s ability to detect motion depending on the direction in which the stimulus is moving relative to the head. For the walking conditions, the visual stimulus was manually triggered by the experimenter with a 1–2 s time window between stimuli regardless of the phase of the gait cycle. For example, the visual stimulus may have rotated to the right while the participant was swaying to the left or right, which would add to or reduce visual motion on the retina, respectively. Controlling the onset of the visual stimulus relative to the gait cycle may reduce the variability of thresholds measured in the current investigation.

Beyond accurate estimates of visual detection thresholds, it may be beneficial to understand if the influence of visual movement on upright balance changes during the gait cycle. For example, during double stance, more information is available from lower limb proprioceptors and thus may lead to less reliance on vision. In contrast, during single stance, in which the contralateral leg is in swing phase, visual cues may be more important to maintain balance since there is less contact with the body to the ground. In fact, phase-dependent visual coupling has been observed during walking (Logan et al., 2010). One major mechanism for balance control is modulation of the foot placement based on the state of the body at mid-stance (Wang and Srinivasan, 2014), and visual motion detection is likely used to estimate the body state (i.e., CoM position and velocity).

The influence of age on visual motion thresholds during standing and walking is under studied. Previous literature has indicated an increase in visual motion detection thresholds for older adults compared to young adults (Warren et al., 1989; Gilmore et al., 1992; Tran et al., 1998; Habak and Faubert, 2000; Snowden and Kavanagh, 2006; Conlon et al., 2017), but there are no studies that attempt to measure thresholds in older adults during standing and walking. Older adults are known to place more emphasis on vision (i.e., upweight) while standing (Haibach et al., 2009) and walking (Franz et al., 2015). If motion thresholds are higher for older adults during standing and walking, combined with their higher reliance on vision for upright balance control, larger thresholds may contribute to their fall risk. Saftari and Kwon (2018) emphasize that particularly in the aging literature, the relationship between visual detection thresholds and fall-risk remains a critical knowledge gap.

Overall, our results indicate that a visual motion detection threshold can be reliably measured during walking and standing, and that thresholds are higher for walking than standing. The relationship between visual processing and fall risk has focused on aspects of visual acuity, although not in the case of optic flow detection. The ability to reliably measure a visual motion detection threshold while standing and walking adds to our understanding of visual motion processing and balance control, particularly in populations with higher fall risk.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by the University of Delaware IRB, Newark, DE. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

SD and HR contributed to the conception, implementation, and analysis of this study. SD ran the study and drafted the original manuscript. HR and JJ reviewed and edited the manuscript for publication. All authors contributed to manuscript revision, read, and approved the submitted version.