2.1.1.

Participants

Participants were 14 healthy full-term infants aged 4 to 6 months (6 males; range 4 months 2 days to 6 months 23 days; M $\text{age}=161\text{days}$). Parents reported their infant’s ethnicity/race as Caucasian ($n=11$) or Hispanic ($n=3$). Fifteen additional infants were tested but eliminated from the final sample because of difficulty obtaining an optical signal ($n=11$), procedural error ($n=3$), or crying ($n=2$). Although this number is higher than preferred and higher than our lab typically reports, it is within the range observed in infant fNIRS studies.³⁷ One reason for the high attrition rate in the current experiment is that the headgear covered cortical areas from which it is particularly difficult to measure. The skull overlying occipital–temporal–parietal areas has a significant amount of curvature, along horizontal and vertical planes; it is challenging to get optodes to lay firmly along this curved surface. Parents were offered $5 or a lab T-shirt for participation. This experiment was carried out in accordance with the recommendations and approval (IRB2004-0140D) of the Institutional Review Board, Division of Research, Texas A&M University with written informed consent from the parents/guardians of all infant participants. All parents/guardians gave written informed consent in accordance with the Declaration of Helsinki.

2.1.2.

Apparatus and data recording

A remote eye tracker (Tobii T60 XL) was used to measure eye movements during stimuli presentation. The infrared corneal reflection eye tracker was embedded in the lower portion of a 24 in flat screen monitor set to a resolution of $1024\times 768\text{pixels}$. It detected the position of the pupil and the corneal reflection of the infrared light from both eyes. The Tobii T60 XL records data at 60 Hz with an average accuracy of 0.5 deg visual angle and a head movement compensation drift of G0.1. Fixation data were defined using the Tobii fixation filter (version 2.2.8) with a velocity threshold of 35 pixels and a distance threshold of 35 pixels. The total duration of looking during each test trial was calculated by the fixation data sums for each trial. The monitor was mounted on an adjustable arm so that it could be positioned optimally for each infant. A full-face view of the infant was recorded by a Logitech Webcam Pro 9000 positioned atop the monitor during stimuli presentation. Tobii Studio was used to present the stimuli on a Dell Precision M6400 desktop computer with a Windows XP operating system.

One might be concerned that the wavelength of light emitted by the Tobii eye tracker would interfere with the fNIRS signal. Although Tobii does not publically release specific wavelength measurements for its eye-trackers, the light emitted is within the range of 600 to 900 nm. Hence, the wavelength of infrared light emitted by the eye tracker could potentially overlap with the wavelengths detected by our optical imaging device. We were aware of this potential problem when designing the experiment. We addressed this concern by comparing the fNIRS signal obtained when the eye tracker was off instead of when it was on. No significant change in the signal (off versus on) was observed. As an added precaution, however, once the headgear was placed on the infants head the optode arrays were covered with black, light-blocking material.

In addition to number of fixations (fixation count) and duration of looking, two other visual scanning measures were extracted. Both measures were computed using fixation coordinates $(x,y)$ for each look within each trial. The first measure we calculated was the distance from one fixation coordinate to the next (in pixels); we call this distance. The second measure we calculated was the angle between one fixation coordinate and the next (in deg); we call this measure direction. We reasoned that if infants perceived the 3-D SFM shape, they would scan the moving contour of the shape in a systematic fashion, and this would be associated with HbO responses. In contrast, we predicted that viewing RM displays that do not possess moving contour would not elicit systematically distributed infant scanning. In short, we expected the distance and direction between looks to be smaller (i.e., directed at local contour rather than randomly distributed across the display) when infants viewed the SFM than RM displays.

2.1.3.

Stimuli and design

The SFM and RM stimuli, displayed in Fig. 1, were similar to those of Hirshkowitz and Wilcox (2013) and adapted from Murray et al. (2004). The SFM displays were composed of 450 white dots (against a black background) orthographically projected onto the surfaces of a simple geometric shape (cube or cylinder) that rotated either 90 deg, 135 deg, or 180 deg around the 3-D $y$-axis during each 5-s trial. The RM displays were composed of the same number of dots moving at the same velocity except that the direction of each dots’ motion was randomly assigned. The stimuli (SFM and RM) were $15\mathrm{cm}\times 13\mathrm{cm}$ and presented at the center of the screen. To maintain infants’ attention, each SFM and RM trial was accompanied by a nonverbal, nonmusical sound (e.g., whistle or whirring); sounds were counterbalanced across stimuli. All Infants viewed six pairs of SFM-RM trials (two SFM that each rotated 90 deg, 135 deg, or 180 deg around the 3-D axis), for a total of 12 trials. The order in which the SFM and RM displays were presented within each pair was counterbalanced across participants. Each SFM and RM display was preceded (and followed) by a 10-s display composed of a solid-colored screen that pulsed, with soft music playing in the background. This was used as the baseline display to maintain infants’ attention. We chose not to use a static random-dot scrambled image or a pulsing static random-dot scrambled image to avoid potential dot-shape after effects.

2.1.4.

Procedure

Infants were seated in a parent’s lap 65 cm away from the monitor, in which the stimuli were presented. The testing room was dark and black curtains shielded the infant/parent from the rest of the testing room. Parents were instructed to close their eyes or wear blacked out glasses during the test session. To obtain reliable and valid eye movement data, the Tobii Studio infant calibration program was used prior to stimulus presentation. Animated stimuli were used to direct attention to five gaze positions covering over 80% of the viewing area.

2.1.5.

Functional near-infrared spectroscopy instrumentation

The imaging equipment contained four fiber-optic cables that delivered near-infrared light to the scalp of the participant (emitters), eight fiber-optic cables that detected the diffusely reflected light at the scalp (detectors), and an electronic control box that served as the source of the near-infrared light and the receiver of the reflected light. The control box produced light at wavelengths of 690 nm, which is more sensitive to deoxygenated blood, and 830 nm, which is more sensitive to oxygenated blood, with two laser-emitting diodes (TechEn, Inc.). Laser power emitted from the end of the diode was 4 mW. Light was square wave modulated at audio frequencies of $\sim 4$ to 12 kHz. Each laser had a unique frequency so that synchronous detection could uniquely identify each laser source from the photodetector signal. Ambient illumination from the testing room did not interfere with the laser signals because environmental light sources modulate at a different frequency. Each emitter delivered both wavelengths of light and each detector responded to both wavelengths. The signals received by the electronic control box were recorded and processed via data acquisition software developed by TechEn.

Prior to the experimental session, infants were fitted with a custom-made headgear that secured the fiber optics to the scalp. Configuration of the sources and detectors within the headgear, placement of the sources and detectors on the infant’s head, and location of the corresponding channels are displayed in Fig. 3. The headgear was placed on the infant’s head using O1 and O2 as primary anchors and T3 and T4 as secondary anchors. The headgear was designed to assess hemodynamic responses in dorsal and ventral object processing areas. On the basis of currently available cortical maps for infants^38,39 and adults,⁴⁰ we extrapolated that channels $3/9$ and $5/11$ measured from middle/upper occipital gyrus, including MT+/V5; that channels $1/7$ and $2/8$ measured from occipital–temporal areas, including LOC; that channels $4/10$ channels measured from posterior temporal cortex; and that channels $6/12$ measured from inferior parietal cortex. Source–detector separation was 2 cm. The headgear was not elastic so the distance between sources and detectors on each side of the head remained fixed. The mean head circumference of infants was 42.9 cm ($\mathrm{SD}=1.4\mathrm{cm}$). We also measured the distance between infants’ nasion and inion (N-I) ($M=24.7\mathrm{cm}$, $\mathrm{SD}=2.7\mathrm{cm}$) and between internal auditory canals (IAC–IAC) over the top of the head ($M=24.9\mathrm{cm}$, $\mathrm{SD}=3.1\mathrm{cm}$).

Fig. 3

Configuration of emitters (red circles) and detectors (black squares), and the six corresponding channels (numbered) from which optical signals were measured, in the (a) left and (b) right hemispheres. Approximate location of the emitters and detectors and corresponding channels are placed on a schematic of an infant’s head relative to the 10–20 International EEG system. The circled channels are those from which a significant response to the SFM stimuli was obtained. (c) Infant participating in the study.

2.1.6.

Processing of functional near-infrared spectroscopy data

The fNIRS test data were processed, for each channel and event condition separately, using Homer2. The raw signals were acquired at the rate of 50 samples per second, converted to relative concentrations of oxygenated (HbO) and deoxygenated (HbR) blood using the modified Beer–Lambert law. The optical signals were digitally low-pass filtered at 0.50 Hz and high-pass filtered at 0.010 Hz, and a principal components analysis (PCA) was performed to remove systemic physiology common across channels (accounting for up to 80% of the variance). A recursive motion correction procedure⁴¹ was implemented to objectively identify and replace (if possible) channel specific motion artifacts. A motion artifact was defined as a change in the absolute signal amplitude of greater than 0.5 units over 0.5 s (using a mask of 1 s). A targeted PCA filter removed up to 97% of the variance in the signal due to motion, with a maximum of five iterations for correction. For each test trial, changes in HbO and HbR were examined relative to the 2-s prior to the onset of the test event (i.e., the mean optical signal from $-2$ to 0 was set to 0 and this was used as the baseline). Of the 84 possible SFM and 84 possible RM trials ($14\text{infants}\times 6\text{trials}$ for each condition), infants completed 72 SFM and 78 RM trials. The mean number of trials completed in the SFM condition ($M=4.86$, $\mathrm{SD}=1.29$) and the RM condition ($M=5.14$, $\mathrm{SD}=0.95$) did not differ significantly ($t=1.30$, $df=13$, $p=0.218$).

2.2.1.

Preliminary analyses

Preliminary analysis of the eye tracking (mean duration of looking and fixation count) and optical imaging data revealed no significant effects, or interactions, involving sex. Hence, this factor will not be included in the main analyses. However, given the relatively small number of males and females tested, null effects should be interpreted with caution.

2.2.2.

Eye tracking data

The number of times infants fixated to the test stimuli was averaged over trials and infants for each event condition separately. Paired-sample $t$-tests indicated that fixation counts to the SFM ($M=0.83$, $\mathrm{SD}=0.36$) and the RM ($M=0.92$, $\mathrm{SD}=0.37$) stimuli did not differ significantly ($t=-1.40$, $df=13$, and $p=0.19$). Infants’ duration of looking to the SFM and RM stimuli was also averaged, separately, over trials and infants. Paired-sample $t$-tests indicated that duration of looking to the SFM ($M=2.12$, $\mathrm{SD}=1.70$) and the RM ($M=2.12$, $\mathrm{SD}=1.70$) stimuli did not differ significantly ($t<1$, $df=13$). Together, these data suggest that the infants spent about the same amount of time attending to the SFM and the RM stimuli. Hence, differences in hemodynamic responses to SFM and RM stimuli cannot be attributed to attention factors.

Infants’ mean distance and direction between looks (averaged over trials and infants for each event condition, separately) were analyzed in the same manner as mean fixation and duration of looking. Paired-sample $t$-tests indicated that distance of looks (in pixels) to the SFM ($M=179.58$, $\mathrm{SD}=89.38$) and the RM ($M=210.78$, $\mathrm{SD}=67.03$) stimuli did not differ significantly ($t=-1.25$, $df=13$, $p=0.23$). Paired-sample $t$-tests indicated that direction of looking (in degree of angle) to the SFM ($M=44.60$, $\mathrm{SD}=7.14$) and the RM ($M=44.08$, $\mathrm{SD}=10.66$) stimuli did not differ significantly ($t=1.31$, $df=13$, $p=0.21$).

2.2.3.

Optical imaging data

Hemodynamic response curves for each condition and channel are presented in Fig. 4. For each of the 12 channels (6 channels within each hemisphere), hemodynamic responses were averaged over 5 to 10 s. This interval was chosen because pilot data suggested that hemodynamic responses to the SFM and RM stimuli are well initiated by 5-s poststimulus and return to baseline by 15-s poststimulus. Responses were then averaged over trial and infant, for each of the two event conditions, separately. Mean hemodynamic responses are reported in Table 1 (HbO) and Table 2 (HbR). However, because HbO responses are typically more robust than HbR responses,⁴² we focus our analyses on HbO.

Fig. 4

Mean hemodynamic responses in the left and right hemisphere to the SFM and RM stimuli. O1/O2 and T5/T6 correspond to the International 10–20 coordinates shown in Figs. 3(a) and 3(b). The red curves indicate change in oxyhemoglobin concentration (HbO), the blue curves indicate change in deoxyhemoglobin concentration (HbR), and the green curves indicate the sum total of HbO and HbR (HbT). The black vertical lines indicate time points 0, 5, and 15 s, respectively, (0 and 5 s correspond with the onset and offset of the stimulus). The horizontal axis indicates time ($-2$ to 20 s) and the vertical axis indicates change in optical density units ($\mathrm{\Delta }\mathrm{OD}$, in $\mu \mathrm{M}$ cm). The numbers under each waveform indicate the channel from which the data were obtained. The highlighted channels indicate the channels from which a significant HbO response was obtained.

Table 1

Mean and (standard deviation) HbO responses to the SFM and RM stimuli, for each channel. T and p values (one-tailed) are reported for each channel. BFs are also reported for each channel. Those channels, in which the t-test effect was significant (p<0.05), are highlighted with grayscale. Cohen’s d are reported for channels, in which a response was obtained in the predicted direction.

Table 2

Mean and (standard deviation) HbR responses to the SFM and RM stimuli, for each channel.

Data were analyzed in two steps. First, relative changes in mean HbO for each condition and channel were compared to 0 using $t$-tests (Table 1). We had a directional hypothesis (changes in HbO would be in the positive direction), hence one-tailed tests were used. In the left hemisphere, significant activation was obtained in channel 3. In the right hemisphere, significant activation was obtained in channels 7, 8, and 9. Cohen’s $d$ (see Table 1) indicates medium effect sizes associated with each statistically significant effect.

In addition to conducting $t$-tests, we conducted Bayesian analyses to examine the robustness of our results.⁴³ In these analyses, we assessed the extent to which the alternative hypothesis, an increase in HbO relative to 0 (the null hypothesis would be no increase in HbO relative to 0), was supported. Bayes factors (BFs) are reported in Table 1. A BF indicates that the data obtained are $B$ times more likely under the alternative than null hypothesis. For example, a BF of 3 indicates that the data are 3 times more likely under the alternative hypothesis. A BF between 3 and 10 indicates substantial evidence for the alternative hypothesis. In all four channels, in which significant effects were obtained using $t$-tests, the BF was greater than 3. BFs equal to or greater than 3 were not obtained in any other channels.

In the second data analysis step, we assessed the extent to which the HbO responses obtained to the SFM and RM displays differed significantly from each other. Because the three right hemisphere channels that showed activation to SFM (but not RM) were spatially contiguous, we averaged the responses across those channels to create a right hemisphere region of interest (ROI).^37,44 T-tests revealed that in channel 3 a significantly greater response was obtained to the SFM than RM stimuli, $t=2.53$, $df=13$, $p=0.0125$ (one-tailed), and Cohen’s $d=1.441$. In addition, the mean HbO response obtained in the right hemisphere ROI was significantly greater to the SFM stimuli ($M=0.486$, $\mathrm{SD}=0.661$) than RM stimuli ($M=0.193$, $\mathrm{SD}=0.530$), $t=1.84$, $df=13$, $p=0.044$ (one-tailed), and Cohen’s $d=0.502$. Together, these results provide strong evidence that striate cortex, but not extra striate cortex, responds differentially to SFM than RM stimuli.

2.2.4.

Correlation between visual scanning and HbO responses

To assess the extent to which visual scanning measures were associated with HbO responses to the SFM stimuli (we did not obtain HbO responses to the RM stimuli), we conducted correlational analyses between each of the four eye-tracking measures (fixation count, duration of looking, distance of looks, and direction of looks) and HbO responses in channel 3 and the right hemisphere ROI. The results of these analyses are reported in Table 3. Contrary to our expectations, none of the visual scanning measures was correlated with HbO responses to SFM stimuli in channel 3 or the right hemisphere ROI. Although we have previously reported no difference between SFM and RM for any of these visual scanning measures,⁹ we thought it possible that these measures would be correlated with HbO responses. One interpretation of these data is that hemodynamic responses are a more sensitive measure of SFM processing than behavioral measures.

Table 3

Pearson correlations between the four visual scanning measures (duration of looking, fixation count, distance of looks, direction of looks) and HbO responses to the SFM stimuli obtained in channel 3 and in the right hemisphere ROI (channels 7, 9, and 11). Significance values are two-tailed.