«A Thesis Presented in Partial Fulfillment of the Requirements for the Degree Master of Science Approved April 2014 by the Graduate Supervisory ...»
age (Hindmarch, 1988). Importantly, CFFT is positively correlated with visual experience of directional-motion (Seitz, Watanabe 2003; Seitz et al 2006), intelligence (Halstead 1947), reading speed and comprehension (Groth 2013), and word decoding (Nanez, Holloway, Donahoe, & Seitz, 2006). Moreover, single-cell (Colby, Duhamel, Goldberg, 1993), Seitz and Watanabe (2005); Seitz, Nanez, Holloway, & Watanabe, (2006) and lesion studies (Merigan, Byrne, Maunsell, 1991; Schiller, Logothetis, & Charles, 1991) have indicated that the megnocellaur pathway is particularly sensitive to flicker perception. The magnocelluar pathway, as discussed above, activates to fast moving and low contrast stimuli. In this paper, CFFT was operationalized as general cortical processing or how fast the observer processes information.
The purpose of this experiment was to examine the effects of a performancedependent task consisting of unidirectional-implied-motion on CFFT. First, we evaluated the validity of the implied motion. Second, we examined the validity of the implied motion training. Third, we examined whether performance on implied motion training parallels performance on a real motion task. In the dot motion task paradigm, performance increases as a function of time and practice. Fourth, we examined luminance difficulty and training days on performance. Fifth, we examined the relationship between exposure to implied motion training and CFFT. Based on previous literature, we hypothesis that performance in detecting the direction of implied motion would increase over the four days of training. Second, we hypothesized that enhancement in implied motion training should lead to faster CFFT.
Participants Seven students (Males = 3, Females = 4, age = 18-48 years, M = 26.38, SD = 9.03) were recruited from Arizona State University West Campus. Students were compensated for their time and effort. All participants reported good visual health on a subjective survey and met the 20/40 visual acuity cut off line. Participants signed a consent form and were told that they could quit the experiment at any time. Finally, participants were naïve to the goals of the experiment.
Materials and procedure Stimuli were presented on 24’’ 2407WFP-HC monitor at a resolution of 1920 x 1200 at 76 Hz refresh rate. The experiment was conducted using custom software.
The implied motion stimulus follows the image properties of previous literature (Kourtzi, Kanwisher, 2000; Lorteije et al., 2011). The most frequent visual cues used to imply motion include high articulation in body posture and motion blur. In this study, three software programs were used to control for orientation, luminance, size of stimulus, and visual cues. For example, the shape and the articulation of silhouettes were designed on DAZ Studio 4.6 software. The articulations of the silhouettes varied (categorically) from low speed (little or no implied motion; Figure 3), medium (low implied motion;
Figure 2) and to high (high implied motion; Figure 1). Photoshop Software was used to modify stimulus size (32.4 ° x 18.56°), contrast levels (luminance Y ≈.6,.9, 1.2, 1.9), motion blur (5 pixels at 34°), and orientation (54°, 124°, 270°, 320°).
Implied Motion task
luminance levels, to expose participants with systematic presentation of directional implied motion, and to reduce the effects of habituation. Taken together, it has been shown that all three components (i.e., contrast levels, unidirectional, and reduction of habitation) are needed for effective functional change in visual processing (Kohn, 2007;
Ranganath & Rainer, 2003).
First, participants were presented with white fixation point for 300 ms, followed by a silhouette figure on one of four off-cardinal directions (54°, 124°, 270°, 320°) at different contrast levels (relative luminance Y ≈.6,.9, 1.2, 1.9) for 250 ms. After a silhouette image was presented, a response screen appeared to collect the participants answers. The response screen had four arrows corresponding to the direction of the silhouette. Once the participants responded, an audio and visual feedback was provided.
The feedback included a high-pitched tone coupled with a green “O” symbol for correct response and low-pitch tone coupled with “X” symbol for incorrect response (see Figure 10). The direction of the silhouettes and the contrast levels were randomized across each frame (i.e., trails) and subjects. Finally, the task was designed to increase or decrease in difficulty based on participants’ performance. All participants completed a total of 800 trails per day x four total days of training equaling to 3200 total frames. Software diagnostics were conducted to examine if the correct stimulus were presented, and if the performance-based equation was working properly.
CFFT was measured using a Macular Pigment Optical Densitometer (Wooten, Hammond Jr, Land, & Snodderly, 1999). CFFT was calculated psychophysically by measuring each participant’s sensitivity to a green light (peak wavelength = 550 nm at
in a 1° circle. The experimenter increases the frequency (Hz) of green light (i.e., at equal counter-phase) until the stimulus appears to be a steady light. After this point was established and recorded, the rate of frequency is then increased by 10 Hz, and decreased until the light appears to be flickering again. CFFT was calculated as the average between the frequency at which the light appears to be a steady and the frequency at which the light appears to be flickering. This is measured six times in order to attain an unbiased average for each participant.
The experiment will follow the protocol of Seitz, Nanez, Holloway, & Watanabe, (2006) study. That is, Participants’ head movements were constrained with a chinrest.
The viewing distance of participants was 3 feet away from the monitor. The room was dimly lit at 1.5 cd/m2. Finally, the experiment included a 4 days of implied motion discrimination training, and pre-and post testing of CFFT on day 1 and day 4.
First, we evaluated the efficacy of the implied motion stimulus used in this study.
Previous research has indicated a top-down influence of neural activity on processing of implied motion. That is, there is lag in neural activity when subjects are exposed to an implied motion stimulus. We examined the accuracy of the visual cues using reaction time as the outcome variable. More specifically, we examined the difference in reaction time when participants were exposed to motion blur vs. no motion blur, and when participants were exposed to low, medium, and high degree of articulation. Motion blur refers to streaks surrounding the silhouettes body to imply motion and speed. The results indicated that reaction time to blurred images (M = 840.66 ms) was significantly higher
addition, there was a main effect of degree of articulation on latency, F(2, 24626) = 4.54 p.05) (Table 3). Post hoc analysis using Tukey’s Honestly Significant Difference correction indicated that reaction time for low articulation (M = 796.92 ms) was significantly lower than high articulation (M = 846.66 ms, p.02) (Table 4). No other significant differences were found. As expected reaction time by articulation (Figure 5) and motion (Figure 6) decreased each day, and equalized by day 4. In summary, the visual cues of the stimulus in this study are consistent with the psychophysical and imagining studies of implied motion (Urgesi, Moro, Candidi, & Aglioti, 2006).
Second, we evaluated the validity of the implied motion task software program.
More specifically, we examined the performance-based equation (increase or decrease in difficulty based on observer performance), and proper display of the stimulus. Frequency analyses and a One-way ANOVA were conducted to examine the frequency of stimulus exposure and performance on different levels of luminance, respectively. The One-way ANOVA indicated that there was a significant difference between luminance levels and performance, F (3, 12744) = 914.190, p.001) (Table 5). Furthermore, post-hoc tests using Bonferroni indicated that participants performed at chance level for luminance
0.6Y (M =.50), below chance level for luminance.9 Y (M =.25, p.05), and above chance for high luminance 1.20 Y and 1.90 (M =.57, M =.84, respectively) (Table 6).
Interestingly, participants performed better on sub-luminal-equivalent luminance level (Y≈.6 [below visual awareness)] than they did on the supraliminal-equivalent luminance level (Y≈.9 [at visual awareness]) (M =.25, p.05) (Table 10). This finding may be due to the stimulus properties. That is, participants were able to observe the stimulus at
which increased the probability of choosing the incorrect direction. Possibilities for this finding are presented in the Discussion section below.
Third, we examined whether the performance pattern on the implied motion task parallel the linear trend found in perceptual learning of real motion task. As discussed above, gradual, linear increase over the 4 days of training would suggest the presence of perceptual learning of implied motion. A repeated measure ANOVA was conducted to examine the change in performance over the 4 days training phase of the experiment. The Greenhouse-Geisser correction indicated a significant difference in performance over the four days, F (2.99, 13752.39) = 21.47, p.001, ηp2=.01 (Table 7). Furthermore, polynomial contrast analyses indicated that the difference in performance could be explained by a quadratic function, F(1, 4586) = 41.36, p.001, ηp2=.01 (Table 8). That is, when compared to day 1 (baseline) (M =.53) performance significantly increased in day 2 (M =.56), did not change in day 3 (M =.54), and significantly decreased below threshold in day 4 (M =.48) (Figure 6). These results suggest that the point of saturation (overtraining) occurred after day 2 (Censor & Sagi 2008). According to Censor and Sagi (2008), overtraining or over-exposure to a specific type of stimulus will result in saturation, which should reduce how participants effectively process information.
Fourth, we examined luminance levels and training day as predictors of performance. This analysis was conducted to reduce the chance of committing a Type I error (William, Shadish, Cook, & Campbell, 2002). Two-way analyses of variance were conducted to examine the main effects of luminance, days of training, and the interaction effect between days of training and luminance. The analysis indicated that there is
ηp2=.17) (Table 10). However, there are no significant difference between performance and number of training days (p = ns). The result suggests that partitioning the influence of luminance on performance reduces the significant level of day. In addition, the results indicated an interaction effect between training day and luminance levels, F(9, 12719) = 4.70, p.001, ηp2=.003 (Table 10). Simple contrast analyses (separate one-way ANOVAs) were conducted to probe the interaction effect of number of days x luminance level. The results indicated significant difference in performance for luminance level
1.20Y (F(3, 2987) = 6.29, p.001 and 1.90Y, F(3, 3459) = 6.29, p.001, respectively) (Table 15). Post-hoc analysis using the Bonferroni test indicated that participant’s performance on luminance difficulty of 1.20Y increased significantly from day 1 (M =.53) to day 3 (M =.62, p.01). In addition, participant’s performance on luminance difficulty of 1.90Y increased significantly from day 1 (M =.53) to day 2 (M =.62, p=.03) (Figure 8). Overall, the results provide more evidence for the point of saturation or overtraining occurring after the day 2.
Fifth, we examined the effects of implied motion training on a cortical processing task, namely the subject’s critical fusion function threshold (CFFT). Repeated measures ANOVA were conducted to examine changes in CFFT from day 1 to day 4. The Greenhouse-Geisser correction indicated a significant increase in CFFT score from preimplied motion training (M = 22.23 Hz) to post-implied motion training (M = 24.00 Hz, F(1, 6) = 9.19, p.05, ηp2=.61) (Table 14). This finding corroborates similar findings in numerous studies concerning perceptual learning and plasticity (change in CFFT) to subliminal (Watanabe, Náñéz and Sasaki, 2001) and supraliminal directional motion
supraliminal motion training, implied motion training significantly increases cortical processing, as measured by changes in CFFT (Figure 9), despite the decrease in performance the 4 days of training. Note that we did not compare the changes in CFFT to a control group. Previous research has shown that CFFT is stable measurement for visually healthy participants (Seitz, Nanez, Holloway, & Watanabe, 2006). Furthermore, in a similar study, Seitz al., 2006 has shown that CFFT does not increase for n-back (black frame), flash of dots (no motion), and control groups. We reasoned that the withinsubject design with many repeated trials over time (four days in the current study) is sufficient to conclude that the changes in CFFT were due to the implied motion task (William, Shadish, Cook, & Campbell, 2002).
The goal of this experiment was to expand on previous research that has demonstrated a direct link between low-level visual processing of directional-motion and cortical processing (e.g., flicker perception). A secondary goal was to expand on fMRI studies of implied motion, through psychophysical experimentation. In this study, we examined the effects of an implied motion discrimination task on cortical processing (CFFT). The rationale for examining implied motion as a predictor of cortical processing is based on studies that show a link between implied motion and neural activity in area MT/MST+ (the same area that activates in response to real motion stimuli).
First, we examined whether the stimuli that were designed using custom software in our lab in collaboration with Dr. Aaron Seitz from UC Riverside consisted of valid measures of implied motion. The analysis supported the validity of the stimuli as