Aging Does Not Affect Integration Times for the Perception of Depth from Motion Parallax

Abstract

To successfully navigate throughout the world, observers must rapidly recover depth information. One depth cue that is especially important for a moving observer is motion parallax. To perceive unambiguous depth from motion parallax, the visual system must integrate information from two different proximal signals, retinal image motion and a pursuit eye movement. Previous research has shown that aging affects both of these necessary components for motion parallax depth perception, but no research has yet investigated how aging affects the mechanism for integrating motion and pursuit information to recover depth from motion parallax. The goal of the current experiment was to assess the integration time required by older adults to process depth information. In four psychophysical conditions, younger and older observers made motion and depth judgments about stationary or translating random-dot stimuli. Stimulus presentations in all four psychophysical conditions were followed by a high-contrast pattern mask, and minimum stimulus presentation durations (stimulus-to-mask onset asynchrony, or SOA) were measured. These SOAs reflect the minimum neural processing time required to make motion and motion parallax depth judgments. Pursuit latency was also measured. The results revealed that, after accounting for age-related delays in motion processing and pursuit onset, older and younger adults required similar temporal intervals to combine retinal image motion with an internal pursuit signal for the perception of depth. These results suggest that the mechanism for motion and pursuit integration is not affected by age.

1. Introduction

Older adults have deficits in many visuospatial domains, including motion and depth information processing (Andersen, 2012; Owsley, 2011). Many of these age-related deficits are linked to mobility problems, such as balance issues and falls, and accidents while operating motor vehicles (Choy, Brauer, & Nitz, 2008; Owsley et al., 1998); as such, a full understanding of how visuospatial functioning is affected by age is important for both preventing and alleviating negative outcomes. One visuospatial process that is integral to successfully navigating throughout the world is the recovery of depth from motion parallax.

Motion parallax (MP) is produced through the translation of an observer or scene. During translation, the moving observer maintains fixation on objects within the scene, generating smooth pursuit eye movements (Miles & Busettini, 1992), while stationary objects within the scene appear to move relative to one another, creating relative image motion on the retina. The visual system integrates the motion and pursuit information to generate a depth percept (Nawrot & Joyce, 2006). The Motion/Pursuit Ratio (M/PR) describes the geometric relationship of the velocity of objects moving on the retina (dθ), pursuit eye movement velocity (dα), viewing distance to the point of fixation (f), and object distance from fixation (d) (Nawrot & Stroyan, 2009; Stroyan & Nawrot, 2011):

$$\frac{d}{f}\approx \frac{\mathrm{d}\mathrm{\theta }}{\mathrm{d}\mathrm{\alpha }}$$ (1)

The role of the pursuit eye movement signal is to disambiguate the depth sign in the perception of depth from MP. The relationship of retinal image motion and pursuit eye movement direction is an orderly one: objects with retinal motion in the same direction as pursuit are perceived as being nearer in depth than objects with retinal image motion in the opposite direction (Nawrot & Joyce, 2006). Indeed, an intact pursuit signal is so necessary for disambiguating depth from MP, that experimentally controlling pursuit eye movements—that is, effectively nulling the eye movement that is usually necessary for maintaining fixation on a translating stimulus—results in an ambiguous depth percept (Naji & Freeman, 2004; Nawrot & Joyce, 2006). Neurophysiological studies have likewise shown that the pursuit eye movement signal is necessary for disambiguating depth from MP. Neurons in area MT of the macaque monkey are responsive to depth from motion parallax, and even show depth-sign selectivity (Nadler et al., 2008; Nadler et al., 2009). Interestingly, the responses of these depth selective neurons were tied to the direction of pursuit eye movements, and not to the direction of head movements.

The geometry of depth from MP is, by nature, dynamic. That is, as one moves through the environment, or as objects in the environment move, the parameters of the M/PR (dθ, dα, and f) change in relation to one another and to the observer. Thus, recovering relative depth information quickly is important for successful navigation throughout the world. Using a masking paradigm, Nawrot and Stroyan (2012) found that younger observers require only 65–75 ms to integrate motion and pursuit signals to recover depth from MP. This rapid recovery of MP depth information is contrary to the results of prior research, which showed that depth from motion is slow and must “build up” (see, e.g., Andersen & Bradley, 1998). A natural (and important) extension of this finding is to investigate the temporal parameters of depth from MP in older adults. Recent research has shown that age independently affects the motion and pursuit signals for depth from MP (Holmin & Nawrot, 2016). Motion thresholds increase and pursuit accuracy decreases with increasing age, resulting in different motion and pursuit signals in younger and older adults.

Other research has examined the effects of age on the temporal parameters for integrating motion information and for initiating pursuit eye movements—that is, generating or acquiring the component signals for depth from MP. The few published studies of aging and temporal integration of motion information have produced conflicting results. Roudaia and colleagues (2010) presented younger and older observers with a random-dot two-frame apparent motion sequence in which the magnitude of dot displacement and the duration of the interstimulus interval (ISI) varied. Observers were required to discriminate the direction of dot displacement. At higher ISIs (60–160 ms), older observers performed worse (i.e., made fewer correct direction judgments) than younger adults across all displacement levels. That is, the maximum temporal interval at which two frames could be integrated in order to make a directional judgment was reduced in older observers. The results of Roudaia et al. indicate, then, that older adults have deficits in temporal integration of motion information.

In contrast to Roudaia et al. (2010), in a study of shape identification, Andersen and Ni (2008) found no effect of age on temporal integration. In their first study, older and younger adults identified two-dimensional (2-D) shapes defined by spatial and temporal properties. An opaque shape, such as a triangle, was drawn on a random-dot background. The shape had no boundaries, but translated across the background, occluding background dots as it translated. This “accretion and deletion” of the background texture information allowed the boundary of the shape to be identified. When the dot density was decreased, older observers’ shape identification performance was worse than younger adults’, indicating that spatial integration was impaired in older adults. By manipulating the velocity at which the shape translated, the amount of information that was available to observers was increased or decreased: increasing velocity increased the rate of accretion/deletion of background elements, thereby providing more temporal information. Both younger and older adults’ performance increased as velocity increased, and the rate of increase across velocity was constant for younger and older adults, suggesting that age does not affect temporal integration. In a second study, spatial information (density) and velocity (8 deg/sec translation) were held constant, while the individual dots that made up the background were varied by point lifetimes (i.e., duration of the individual dots). Again, there was no effect of age on performance—older and younger adults performed equivalently across different point lifetimes. Similarly, Arena, Hutchinson, and Shimozaki (2012) found that there was no effect of age on global motion thresholds when the dots comprising the stimulus were varied in speed, indicating that age did not affect temporal integration.

While the results of these studies appear at first to be contradictory, there are several conceptual differences that must be taken into consideration. Roudaia et al.’s task essentially assessed observers’ abilities to integrate motion information over two frames (i.e., displacement thresholds), while Andersen and Ni’s and Arena et al.’s studies assessed performance for continuously-moving stimuli that varied in velocity or point lifetimes. It is therefore likely that these studies measure different aspects of temporal integration (velocity vs. point lifetimes vs. displacement) in older adults. Another important difference in these studies is that in Roudaia et al.’s study, the longest stimulus duration was 440 ms, while the stimulus duration in Arena et al.’s study was 853 ms, and was 5 sec in Andersen and Ni’s. It is possible that older adults’ temporal processing of motion information is slowed, but that this age effect will not be apparent given a stimulus of prolonged duration, hence the conflicting results across these three studies.

It is not only motion processing (dθ) that is affected by age. Studies of the effects of age on pursuit eye movements have revealed that older adults require longer temporal intervals to initiate a pursuit eye movement. Older adults typically have pursuit latencies approximately 35–50 ms longer than younger adults’, for pursuit stimuli translating at velocities between 5 and 20 deg/sec (Knox, Davidson, & Anderson, 2005; Sharpe & Sylvester, 1978). Handke and Büttner (1999) also found a significant age difference in pursuit onset for a target translating at 10 deg/sec; however, older adults had only 10 ms longer latencies compared to younger adults.

In summary, evidence suggests that older adults are delayed in processing motion and pursuit information, and, by extension, take longer to generate or acquire the component dθ and dα signals necessary for depth from MP. In addition to generating these component signals, the visual system must also integrate the two signals in order to produce the perception of depth from MP. Younger adults require approximately 20–35 ms to recover motion information, and 65–75 ms to recover depth information (Nawrot & Stroyan, 2012). The additional 40–45 ms necessary for younger adults to complete MP depth processing (that is, the processing interval beyond the 20–35 ms motion processing time) likely reflects the processing time necessary for acquiring and integrating the pursuit (dα) signal with the motion (dθ) signal to generate a perception of depth from MP. The goal of the current experiment was to assess the time required by older adults to integrate motion and pursuit to generate a depth percept. Considering older adults’ delayed processing for motion and pursuit information, it is possible that the temporal parameters of the motion/pursuit integration mechanism will likewise be delayed, reflecting generalized slowing of visuospatial processing (Salthouse, 1996). Alternatively, aging may have no effect on the temporal parameters of the integration mechanism, and any slowing in MP depth processing will reflect slowing in the motion and pursuit processes independently.

In the current experiment, observers made judgments about motion direction or about depth phase across four different conditions. To assess processing delays, we employed a backward masking paradigm to measure threshold stimulus durations necessary for younger and older observers to make motion and depth judgments. In masking paradigms, in order to make a stimulus judgment, all processing must occur before the appearance of the mask, which interrupts stimulus processing (Breitmeyer, 1980; Enns & Di Lollo, 2000). In no-mask conditions, stimulus processing may continue after the stimulus has been removed, so that observers may require only very brief stimulus presentations in order to make a subsequent stimulus judgment. As such, threshold stimulus durations are generally shorter in no-mask conditions (Nawrot & Stroyan, 2012), but the duration thresholds found in masking conditions may better reflect the temporal parameters of the neural mechanisms underlying stimulus processing (Roinishvili et al., 2011). Therefore, in the current study we defined threshold stimulus duration as the stimulus-to-mask onset, or SOA. Threshold SOAs in each task were found using an adaptive procedure. Pursuit latency and accuracy (gain) for a translating target (4 deg/sec) were also measured.

2. Material and Methods

2.1 Observers

Eight younger (age range: 18–21 years, M = 20.3, SD = 1.1) and 16 older (age range: 61–73 years, M = 66.6, SD = 4.3) observers participated in the current study. The mechanisms of MP integration in younger adults have been well-described in previous studies (see Lauer & Nawrot, 2015; Nawrot & Stroyan, 2012); therefore, the focus of the current experiment was on describing MP processes in older adults. As such, a greater number of older adults than younger adults participated in the current study. However, as the reader will see in the Results section, the effect sizes are quite large, indicating that any loss of power due to unequal Ns was minimal.

Monocular and binocular acuity was assessed with a 10 foot Graham-Field acuity chart. All observers had normal or corrected-to-normal vision (20/25 or better), and gave informed consent to participate. Only older observers with no self-reported ocular diseases were eligible to participate. Observers’ neurological health status was assessed using a revised form of Christensen, Armson, Moye, and Kern’s (1992) health questionnaire, and observers with self-reported neurological diseases were excluded from participating. Younger observers were volunteers or were given course credit for their participation. Older observers were compensated $10/hour for their participation. The procedures were overseen by the local Institutional Review Board and in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki).

2.2 Apparatus

Stimuli were generated on a Macintosh computer and presented on a 21” flat screen NEC CRT monitor with a 1024 × 768 pixel resolution and 120 hz refresh rate (8.3 ms frame duration). An ASL Eye-trac 6000 (Applied Science Laboratory, Bedford, MA) with remote optics and a sampling rate of 120 Hz was used to measure eye position. The eye tracking system is accurate within 0.5 deg, with a precision of less than 0.25 deg. The system communicated eye position information to the stimulus computer through a 16-bit analog connection with a National Instruments multifunction I/O board. Observers used a keyboard to initiate trials and record responses during the psychophysical task.

Experiments were conducted in a dimly-lit (~ 1 lux) room, at a viewing distance of 57 cm. At this viewing distance, the display subtended 34.33 deg, and each pixel subtended 2.3 minarc. Observer movement was restricted by a chinrest, and an eye patch occluded one of the observers’ eyes for monocular viewing, so that observers were using their preferred eye during the experiment.

2.3 Stimuli & Procedures

The design of the psychophysical stimuli in the current experiment is detailed in Nawrot and Stroyan (2012), and is described briefly below. In the pursuit task, the pursuit target was a small (0.23 × 0.23 deg) white target that translated leftward or rightward at 4.6 deg/sec. All observers completed the pursuit task before completing the psychophysical conditions. Data collection began with a 9-point calibration of the ASL eye tracking system, followed by a 2-point calibration of the experimental computer’s recording of the eye position signal, and a final 5-point calibration along the horizontal axis of the pursuit target’s movement. The experimenter initiated a pursuit trial by button press, after ensuring that the observers were fixating the central target. Initiation of a trial caused the pursuit stimulus to “step” either to the left or to the right (Rashbass, 1961), and then begin translating across the screen in the opposite direction. The amplitude of the step was such that for each trial the translating pursuit target passed through the original fixation spot 100 ms after onset of translation. The target was erased from the display at 606 ms and eye position was recorded for an additional 224 ms. The observers’ task was to maintain fixation on the target as it translated. All observers completed three blocks of trials, with four trials (two leftward translation, two rightward translation) in each block. The purpose of the pursuit task was not only to measure pursuit latency and accuracy, but also to ensure that all observers could generate and maintain pursuit eye movements at the stimulus velocity used in the psychophysical conditions.

In all four psychophysical conditions, the stimuli were comprised of 4,000 2.3 × 2.3 min white dots contained within a 6.6 × 6.6 deg stimulus window. The procedure for each psychophysical condition is described in detail below. Generally, for all four conditions, each trial began with a fixation point centered on the display. Following a button press by the observers, the fixation square either stepped laterally to the left or right (in the depth conditions), or remained in the center of the screen (in the motion conditions). Following a variable 750–1500 ms fixation interval, the stimulus window was drawn centered on the fixation point. During all four conditions, the observers were instructed to maintain their gaze on the fixation point, which remained centered within the stimulus window throughout stimulus translation. Fixation was enforced by the eye tracker within a 1 deg window around the fixation point. For each trial, stimulus presentation was dependent upon fixation within this window, and trials in which fixation was lost were discarded and the trials were immediately repeated.

A high-contrast pattern mask immediately followed the stimulus presentation (0 ISI), thus interrupting stimulus processing (Breitmeyer, 1980). The mask was composed of 50 black and 50 white circles of varying sizes drawn on a 13 × 13 deg black background. The largest circle subtended 7 deg, and the smallest, 2 deg. Alternating by contrast, the circles were drawn in random positions on the background. Several different mask stimuli were created for each block of trials. After the mask was removed, observers entered their response (a perceptual judgment, described below) by button press. The stimuli were presented in two interleaved staircases in which the stimulus presentation duration was varied for leftward and rightward stimulus translations. For both the leftward and rightward staircases, stimulus duration started at 167 ms (20 frames) and descended in 42 ms steps (five frames) until the first reversal or the 42 ms total presentation duration, and stepped in 8.3 ms intervals (one frame) subsequently. The staircases used a three-down, one-up decision rule (Wetherill & Levitt, 1965), and ended at six reversals.

2.4.1 Depth conditions

In two depth conditions, the entire stimulus window translated at 4.6 deg/sec, and the dots within the stimulus window translated horizontally with a sinusoidal velocity profile, with a peak translation velocity of 2.0 deg/sec. The combination of dot translation (dθ) and stimulus window translation (dα) produced a stimulus that appeared corrugated in depth, with 0.75 cycle/deg above and 0.75 cycle/deg below the central fixation point. The depth phase of the stimulus was determined by the direction of dot translation (dθ) in relation to the direction of window translation (dα). Dots that translated in the same direction as the window translation were perceived as near in depth.

Observers completed one block of trials in each depth condition. In both conditions, observers made a judgment about the depth phase of the stimulus by indicating the half-cycle of the stimulus that appeared “far” in depth, relative to the fixation point. The goal was to measure threshold stimulus durations (stimulus-to-mask onset asynchrony, or SOAs) necessary for observers to make accurate depth phase judgments without interference from the high-contrast pattern mask. Observers maintained their gaze on the fixation point throughout stimulus presentation.

2.4.1.1 Condition 1: depth without pursuit prelude

Following a variable 750–1500 ms fixation interval, the MP stimulus was drawn, centered on the fixation point, and immediately began translating laterally at 4.6 deg/sec. The magnitude of the fixation step was calculated so that the stimulus window traversed the center of the display halfway through its translation period.

2.4.1.2 Condition 2: depth with pursuit prelude

Following a variable 750–1500 ms fixation interval, the fixation square commenced translating (4.6 deg/sec) toward the center of the screen. The fixation square translated for a variable interval between 750–1500 ms; following this interval, the MP stimulus window was drawn, centered on the translating fixation point. The stimulus window and fixation point continued translating at 4.6 deg/sec. For each trial, based on the pursuit prelude interval, the magnitude of the initial fixation step was calculated so that the MP stimulus window traversed the center of the display halfway through its translation period.

2.4.2 Motion conditions

In the motion conditions, the stimulus parameters were the same (i.e., in number of dots, size) as in the depth condition, except that these stimuli did not contain depth information.

Observers completed one block of trials in each motion condition. In both conditions, observers made a judgment about the direction of stimulus motion. The goal was to measure threshold stimulus durations (SOAs) necessary for observers to make accurate motion direction judgments without interference from the high-contrast pattern mask.

2.4.2.1 Condition 3: stimulus relative motion

Following a variable 750–1500 ms fixation interval, the stimulus window was drawn, centered on the fixation point. The stimulus window remained stationary, while the dots within the stimulus window translated (2 deg/sec): dots in the upper and lower thirds of the stimulus translated laterally in one direction (either leftwards or rightwards), and the dots in the middle third of the stimulus translated in the opposite direction, creating relative image motion. Observers indicated the direction of motion of the dots in the center region of the stimulus window.

2.4.2.2 Condition 4: stimulus window motion

Following a variable 750–1500 ms fixation interval, the stimulus window was drawn, centered on the fixation point. The stimulus window immediately began translating leftward or rightward (4.6 deg/sec). The dots within the stimulus window remained stationary. Observers indicated the direction of motion of the stimulus window.

3. Results

Analyses were conducted in MATLAB 2014b (MathWorks, Natick, MA), Microsoft Excel 2011, and SPSS 21 (SPSS II, New York, NY). Initial analyses showed that there was no difference between leftward- and rightward-translating stimuli in any of the measures (two depth conditions, two motion conditions, and pursuit) for either of the two age groups (main effect of direction: F(1,22) < 1, p > 0.05; direction × age interaction: F(1,22) < 1, p > 0.05). Therefore, data were collapsed over direction for all further analyses.

MATLAB was used for all eye-tracking analyses. Eye position data were smoothed using a 10-term finite impulse response filter (see Carl & Gellman, 1987, for similar filtering methods), and eye velocity was derived from eye position using a two-point central difference algorithm. Data were then low-pass filtered at 40 hz using a 3-term moving average filter. Saccades were identified as velocities greater than 40 deg/sec (Burke & Barnes, 2006) and those sections were removed from further analysis.

To analyze gain, the first 150 ms and the last 224 ms of the velocity data were discarded, to exclude the pursuit initiation phase and pursuit after the target had disappeared from the screen. Gains were calculated by averaging velocity over the remaining data points (456 ms) and computing the average eye velocity/target velocity. A one-way ANOVA revealed a significant difference in younger (M = 1.00, SD = 0.1) and older (M = 1.11, SD = 0.1) adults’ gains (F(1,23) = 6.27, p = 0.02).

Pursuit onset was defined as the moment at which eye velocity exceeded 2.3 deg/sec (i.e., 50% of target velocity). The MATLAB analysis program began to search for pursuit onset 80 ms after target onset for younger adults, and 120 ms after target onset for older adults. Constraining the temporal windows for identification of pursuit latency allowed for automated elimination of early, small-amplitude high-frequency eye movements. (Note that there is little consensus on how pursuit onset is defined—some researchers define onset simply as the moment eye velocity becomes greater than zero (Handke & Büttner, 1999; Sharpe & Sylvester, 1975), while others use regression techniques to estimate pursuit onset, which allows for the exclusion of pre-onset noise (Knox et al., 2005). The definition of onset used here likewise allows for this noise exclusion by constraining the temporal window for onset.) Younger adults’ pursuit latencies (M = 147.1 ms, SD = 13.4) were significantly shorter than older adults’ (M = 173.7 ms, SD = 24.1), F(1,23) = 8.36, p < 0.01, η_p² = 0.28.

Average SOAs (stimulus-onset-asynchrony, or stimulus-to-mask durations) for each psychophysical condition were calculated for each observer from the last four reversals of the staircase. As shown in Figure 1, older adults had longer SOAs than younger adults across all four conditions. Table 1 shows the mean SOAs for older adults (OAs, second column), younger adults (YAs, third column), and the age difference in SOA (younger adults’ SOAs subtracted from older adults’ SOAs, last column), for all four psychophysical conditions (standard deviations in parentheses). Older adults required ~50–55 ms more processing time than younger adults to recover motion information. In the depth tasks, older adults’ SOAs were ~75–90 ms longer than younger adults’. A 2 (age group) × 4 (condition) mixed factorial ANOVA confirmed that the age difference was significant across all four tasks (F(3,66) = 73.5, p < 0.001, η_p² = 0.77). The task × age interaction was not significant (F(3,66) = 1.2, p = 0.33). A separate 2 (age group) × 2 (task: motion and MP) mixed factorial ANOVA also revealed a significant effect of task type: SOAs were longer in the two depth conditions than the two motion conditions (F(1,22) = 162.0, p < 0.001).

Figure 1.

Stimulus duration thresholds (SOAs) for younger and older adults across depth and motion conditions. Black bars represent older adults (OA); white bars represent younger adults (YA). Error bars denote standard error.

Table 1.

SOAs for Older and Younger Adults (ms)

Psychophysical conditions	OA	YA	OA-YA
MP pursuit prelude	249.4 (59.8)	160.3 (54.8)	89.1
MP no pursuit prelude	193.7 (38.2)	116.9 (79.7)	76.8
Stimulus window motion	78.5 (34.0)	23.6 (11.6)	54.9
Stimulus relative motion	91.6 (35.7)	41.6 (11.6)	50.0

Bivariate correlations were performed to assess the relationships of age, motion, MP, and pursuit latency. Pearson’s r’s for each relationship are shown in Table 2. Age was significantly positively correlated with all stimulus conditions, indicating that as age increases, motion and depth processing times and pursuit onset times likewise increase. SOAs for both motion conditions (stimulus window motion and stimulus relative motion) and pursuit latency were positively correlated with SOAs in the depth with no pursuit prelude condition—that is, as motion processing time and pursuit onset time increase, the amount of time necessary to process depth in this condition also increases. The two MP conditions are significantly correlated; however, pursuit latency and SOAs in the stimulus relative motion condition are not related to processing times for MP stimuli preceded by a pursuit prelude. The implication of this result will be discussed in Section 4.

Table 2.

Correlation Matrix

Measures	(1)	(2)	(3)	(4)	(5)	(6)
(1) Age	1.00
(2) MP pursuit prelude	0.54**	1.00
(3) MP no pursuit prelude	0.54**	0.53**	1.00
(4) Stimulus window motion	0.71**	0.62**	0.61**	1.00
(5) Stimulus relative motion	0.69**	0.37	0.52**	0.74**	1.00
(6) Pursuit latency	0.53**	0.27	0.41*	0.62**	0.43*	1.00

Note:

^*p < 0.05;

^**p < 0.01

4. Discussion

Overall, compared to younger adults, older adults required longer stimulus presentations to recover MP depth sign information. For younger adults, stimulus presentations as brief as ~115 ms (when stimuli were followed by a high-contrast mask) were sufficient for the observers to recover depth sign information. In contrast, older adults required at least 190 ms. There was also a significant age difference in the motion conditions, in which younger observers required only ~20–40 ms stimulus presentations, but older observers required ~75–90 ms SOAs. This difference in the temporal parameters for depth from MP between the two age groups is not due to the inability of older observers to accurately maintain fixation on the MP stimulus, as all older observers were able to complete the depth tasks with fixation enforced by an eye tracker. It is also important to note that in most psychophysical conditions (all but the MP pursuit prelude condition), stimulus presentations were so brief that no eye movements were initiated before the stimulus disappeared. This suggests that a pursuit eye movement signal, which is generated before the eye begins to move, is all that is required for the perception of depth from MP. That is, younger adults’ pursuit latencies were approximately 147 ms, and older adults’ latencies were about 174 ms; for both age groups, pursuit onset was notably slower than the mean stimulus duration in three of the four psychophysical conditions. Therefore, although there was a significant age difference in pursuit gains, it is unlikely that pursuit accuracy influenced the minimum motion or depth judgment intervals. There was also a significant difference in pursuit latencies: older adults were ~27 ms slower than younger adults in initiating pursuit. This age-related slowing in pursuit onset is similar in magnitude to the effects of age on latency that have been revealed by previous research (Knox et al., 2005; Sharpe & Sylvester, 1978).

As noted in Section 1, the processing time necessary to recover depth from MP is a product of the motion (dθ) processing time, the time required to acquire the pursuit (dα) signal, and the processing interval necessary to integrate these two components to generate a depth percept (Nawrot & Stroyan, 2012). Figure 2 shows how these mechanisms might interact to generate the perception of depth from MP. Previous research (Holmin & Nawrot, 2016) and the results of the current study have shown that older adults have deficits in both of the component processes (motion processing and pursuit). The goal of the current study was to determine if older adults also have deficits in the motion/pursuit integration mechanism.

Figure 2.

Mechanisms of the perception of depth from MP. Two proximal signals, motion and pursuit (left side of the figure), may be additive; for example, a slowed motion processing system may feed into a slowed pursuit planning system (indicated by dashed arrow). The two input signals must be integrated by the visual system (center). After this integration, an unambiguous depth percept is formed (right).

The results shown in Figure 1 suggest that age does affect integration times for depth from MP—older adults’ processing times in both depth conditions are ~80 ms longer than younger adults’ (see Figure 1). One possible interpretation of these results is that the age-related differences in the two inputs to the integration mechanism (motion and pursuit signals) do not directly affect the timing of MP integration, but rather that the differences in MP integration times are caused by age-related slowing of the MP integration mechanism itself. However, a more likely explanation, given the data, is that these longer depth-processing times are actually caused by age differences in motion processing and pursuit onset. Older adults’ motion processing times are slower than younger adults’, by ~50 ms. Pursuit latency is also longer in older adults, by ~30 ms. If age differences in motion processing intervals and pursuit onset are somehow additive (e.g., a slowed motion processing system feeds in to a slowed pursuit planning system), and this additive processing is taken into consideration, older adults’ processing times for depth from MP are very similar to those of younger adults’. Figure 3 demonstrates this relationship of delayed motion processing and pursuit onset. (Figure 3 is adapted from the “MP no pursuit prelude” results in Figure 1; however, the same analysis applies for both MP conditions.) Altogether, the results of the current study suggest that the mechanism of motion and pursuit signal integration is not affected by age. Although older adults do require longer intervals to recover MP depth information, this age difference appears to be caused by age-related slowing in the two front-end mechanisms, motion processing and pursuit, which may interact additively as suggested by the dashed arrow in Figure 2. Unfortunately, it is difficult to disassociate the temporal processing of depth from MP from motion and pursuit processing—by definition, a stimulus that induces the perception of depth from MP contains internal stimulus motion and also generates a pursuit eye movement signal. Removing either of these components from a MP stimulus will also eliminate the perception of depth. The approach we have taken in the current study—to measure motion integration intervals and pursuit onset in isolation and extrapolate to MP integration times—is the most tenable means of investigating temporal integration for depth from MP. Future research might focus on disentangling the contributions of motion processing and pursuit by looking at individual age-related changes in the two components and in MP depth perception within a larger sample of older adults.

Figure 3.

The mechanism of motion/pursuit integration is not affected by age. Age effects on motion processing and pursuit onset (top two brackets) can account for age-related slowing of MP depth processing (bottom bracket). When these age effects are accounted for, older adults’ MP processing time is comparable to younger adults’.

Recall that Andersen and Ni (2008) and Arena et al. (2012) did not find any age differences in their studies of temporal integration, whereas the results of the current study reveal a significant age difference in the temporal parameters for processing motion information. The differences in the results from the current study and previous research may be attributed to differences in how temporal information was varied—Andersen and Ni and Arena et al. varied temporal information by manipulating dot velocity and dot lifetimes, while in the current study, velocity and dot lifetimes were held constant, and stimulus duration was manipulated. The results of the current and previous studies are not irreconcilable; rather, it is likely that different aspects of temporal integration are being measured, as alluded to in Section 1 (and see Roudaia et al., 2010). Given a long or unlimited stimulus duration (as in Andersen & Ni, or Arena et al.), older adults may demonstrate no deficits in motion processing, compared to younger adults; however, temporal processing of motion information is slowed in older compared to younger adults.

Similar to previous results from Nawrot and Stroyan’s (2012) study, observers in both age groups had longer SOAs in the MP with pursuit prelude condition than in the MP with no prelude condition. The reason for the longer processing time in the prelude condition is not immediately clear, but these results provide some insight in to how the visual system may generate and use pursuit (dα) information. In the pursuit prelude condition, the fixation point translates for a variable amount of time before the stimulus appears. The fixation translation interval (750–1500 ms) is long enough so that observers are engaged in closed-loop pursuit when the stimulus appears and they are required to make a depth judgment. It is thought that the oculomotor system has a “velocity storage” mechanism that underlies predictive and anticipatory pursuit (Barnes & Donelan, 1999); if observers were using stored velocity information to generate the dα signal necessary to recover depth from MP, observers might have immediate access to the requisite dα signals, and the threshold SOAs would be lower in the pursuit prelude than the no prelude condition. However, SOAs are longer in the former condition, suggesting that observers cannot use stored velocity information to generate the requisite dα signal—rather, observers use an “active” dα signal generated with the appearance of the MP stimulus. Perhaps, as Nawrot and Stroyan (2012) suggest, the dα signal must be updated in the presence of a new stimulus (i.e., retinal image motion), in order to prevent a perceptual error.

Moreover, it is possible that pursuit signals (dα) are generated differently in the open-loop (initiation) and closed-loop (maintenance) phases of pursuit. A “slower” closed-loop signal would explain why the pursuit-prelude actually increased the necessary stimulus presentation duration compared to the case when the observer’s eye was stationary and an open-loop signal is being generated. Additional evidence for this open-loop – closed-loop difference comes from the correlations between conditions shown in Table 2. Notably, pursuit latency was not significantly correlated with durations in the pursuit prelude condition (r = 0.27, p > 0.05). This suggests that performance in the two tasks reflect different, unrelated, aspects of pursuit. Indeed, pursuit latency indicates the temporal aspects of the open-loop phase of pursuit, while the pursuit prelude condition relies on closed-loop aspects of pursuit. While these results suggest temporal differences the dα signals generated in each of these phases, perhaps there are other differences in the dα signal that would manifest as differences in perceived depth magnitude (Nawrot et al, 2014).

Previous studies have shown that age differences in visual acuity may contribute to differences in older and younger adults’ visual motion processing (Roudaia et al., 2010). It is unlikely, however, that the age differences revealed in the current study were significantly impacted by differences in our observers’ acuity. A number of studies (Ball & Sekuler, 1986; Norman, Ross, Hawkes, & Long, 2003; Trick & Silverman, 1991) have shown that even after manipulating younger adults’ acuities so they “see like” older observers, significant age differences in motion and speed processing remain. Furthermore, it is important to bear in mind that the same stimuli were used in all four psychophysical conditions; consequently, any effects of acuity on motion and MP processing would be consistent across all conditions, and would therefore not affect our conclusion (i.e., that age differences in MP integration time are accounted for by age differences motion processing and pursuit latency).

The slowed motion processing and pursuit onset times in older adults that are found in the current study are consistent with the results of previous studies of age effects on these processes. There is a rich literature documenting higher motion coherence thresholds in older adults, as well as longer motion duration thresholds (Ball & Sekuler, 1986; Bennett, Sekuler, & Sekuler, 2007; Betts et al., 2005). It has been suggested that age-related changes in the transmission of the neurotransmitter γ-aminobutyric acid (GABA), the main inhibitory transmitter in the central nervous system (CNS), underlie age deficits in many visual functions, including motion perception (Leventhal et al., 2003; Schmolesky et al., 2000). Specifically, neurons in old monkey area MT (the region of the brain most often associated with global motion processing) are noisier and have larger tuning curves than young monkey MT neurons, leading to a loss of direction and speed specificity (Liang et al., 2010; Yang et al., 2009). Likewise, a loss of GABA transmission with age may contribute to slowed pursuit onsets in older adults—the onset of pursuit requires a visual motion signal, which likely arises in area MT. Because older adults’ sensitivity to motion is reduced, it follows that pursuit initiation should also be reduced in older adults, as motion input drives pursuit initiation. Noise in MT neurons may also contribute to slowed pursuit initiation latency in older adults (Niu & Lisberger, 2011).

An alternative explanation for these results is that age differences in MP integration times are due solely to age differences in processing of vertical perspective information. To date, the effects of age on the perception of vertical perspective have not been documented; until then, it is difficult to suggest that possible age-related changes in perspective processing can account for slowed MP processing times found here. Recent research has shown that, while vertical perspective is an important cue for depth perception, its utility in disambiguating depth from MP is limited in computer-generated displays such as these (George, Johnson, & Nawrot, 2013). Moreover, for very short stimulus durations used here, by necessity, to measure duration thresholds of ~ 200 msec, it is unlikely that a sufficient magnitude of change in vertical perspective information is available to influence perceived depth sign (Nawrot & Stroyan, 2012). Given a large vertical perspective transformation during longer stimulus-viewing conditions, dynamic perspective may influence the perceived depth sign of an MP stimulus (Papathomas, 2002; 2007; Mahar, DeAngelis & Nawrot, 2013). Therefore, the results of the current study support the view that the perception of depth from MP is the product of retinal image motion signals combined with an internal pursuit signal (Nawrot & Stroyan, 2009; Nadler et al., 2009), a process that is independent of any influence exerted by dynamic perspective cues. These results are incompatible with the view that depth from MP is not an independent cue to depth and is due solely to dynamic perspective cues (Rogers, 2016).

Results of a previous study in our lab (see Holmin & Nawrot, 2016) revealed that older adults have higher thresholds for perceiving depth from MP, compared to younger adults. Specifically, for stimuli translating at slow (2.3 deg/sec) and moderate (10.1 deg/sec) velocities, older observers’ thresholds are 1.5–2 times higher than younger observers’. In the previous paper, we provided real-world examples of how higher thresholds in older adults may affect their ability to navigate as effectively as younger adults. In the first example (see pp. 1689–1690), we demonstrated how higher thresholds would affect an observer walking at a rate of 5 km/h, fixating on an object (F) that is 35 m away, orthogonal to the observer’s walking direction. Under these conditions, the observer’s gaze angle (dα) changes at a rate of 2.3 deg/sec. Given at least 3.8 m of depth between the fixated object F and a relatively nearer or farther distractor object (D), a younger observer could make a reliable discrimination about which object is closer. Under the same conditions, an older observer would require 5.6 m of depth between the objects F and D to make a reliable depth judgment. In the second example, we demonstrated how higher thresholds would affect an observer driving at a rate of 50 km/h while the observer is fixating an object F that is 77 m away (dα = 10.1 deg/sec). In this case, a younger observer would require at least 1.54 m of depth difference between F and D to make a reliable depth judgment, whereas older adults would require at least 3.8 m, or double that required by the younger observer.

It is possible that longer MP processing times in older adults contributes to their higher MP thresholds, perhaps by adding noise in to the depth recovery system. However, it is unlikely that this contribution can account for much of the age difference. Consider, again, an observer walking at 5 km/h. A younger observer will translate 0.16 m in 115 ms (the younger adult SOA in the MP no pursuit prelude condition), creating a change in viewing distance to fixated object F from 35 to 34.84 m. Under the same conditions, an older observer will translate 0.27 m in 190 ms (the older adult SOA in the MP no prelude condition), creating a change in viewing distance to object F from 35 to 34.73 m. The movement of objects F and D in relation to each other will be negligible in such a brief amount of time. The same analysis can be applied to the situation in which an observer is driving at 50 km/h. A younger observer will translate 1.6 m in 115 ms, creating a change in viewing distance from 77 to 75.4 m; an older observer will translate 2.7 m in 190 ms, creating a change in viewing distance from 77 to 74.3 m. As in the walking situation, the relative motion of F and D will be negligible in such brief intervals.

Although the effects of age on MP processing intervals may not contribute substantially to age differences in MP depth thresholds, it is undeniable that rapid recovery of motion and depth information is crucial for effective navigation throughout the world, and slowed processing of this information may contribute to adverse outcomes, such as falls or driving accidents. As the population of older adults in the United States rises rapidly over the next few decades (Colby & Ortman, 2015), a full understanding of the effects of age on visuospatial functions will be important for the treatment and prevention of these negative outcomes.

Highlights.

• Aging affects temporal integration of motion information

• Pursuit eye movement onset is slowed in older adults

• Older adults have slowed processing times for depth from motion parallax

• Age does not affect motion and pursuit integration for depth from motion parallax