Twenty-four infants (15 males, 9 females), who ranged in age from 98 to 125 days old (M = 104, SD = 11), participated in this experiment. They were pseudorandomly assigned to one of three independent experimental conditions (n = 8), with the criterion that equal numbers (n = 4) of caesarean-section and vaginally delivered infants were in each condition. The infants in the sample were Caucasian (n = 14), Hispanic (n = 2), Asian (n = 3), East Asian (n = 2), African (n = 2), and other (n = 1), and were primarily of middle to upper socioeconomic status (SES). The infants (12 delivered via caesarean section, 12 delivered vaginally) were all born full-term, within ±2 weeks of the self-reported due dates, and appeared to be in good health, with no apparent visual or neurological abnormalities. The data from 10 additional infants who participated were excluded from the study because of fussiness (n = 5), equipment or software failure (n = 1), or inattentiveness (i.e., infants were disinterested or looked away from the visual field on more than 50 % of trials; n = 4).
Stimuli and apparatus
During both experimental tasks, the babies were laid supine in a specialized crib and viewed stimuli on an LCD monitor, mounted 48 cm above. Between the infant and the monitor was a 30 × 30 cm infrared-reflecting, visible-transmitting mirror that provided the infant with an unobstructed view of the stimuli on the monitor. A remote pan-tilt infrared eyetracking camera (Model 504, Applied Science Laboratories [www.a-s-l.com], Bedford, MA) was also placed overhead (see Fig. 1). Black felt curtains were hung around the crib to limit light entry and reduce distraction.
Using bright-pupil technology, the pan-tilt eyetracker recorded the infants’ eye movements via reflection in the infrared mirror at a temporal resolution of 60 Hz. Diodes on the camera emitted infrared light that reflected off the infrared mirror onto and back off the infant’s retina through the pupil, to produce a backlit white pupil. The infrared light also produced a point reflection on the corneal surface of the eye. Through proprietary software (Applied Science Laboratories), the eye fixation position was calculated from the relation between the corneal reflection and the centroid of the backlit pupil. The eyetracker was initially calibrated by having the infant look at a stimulus (concentric squares) that was presented at known locations on the LCD screen. This calibration was done in order to equate the recorded eyetracker values of eye location to known locations on the screen. All subsequently recorded eyetracker values were filtered through the calibration file to produce measures of eye position data.
Two Dell computers were used during the experimental session. One computer generated and presented the stimuli using the program DirectRT (Empirisoft Inc., New York, NY; www.empirisoft.com/DirectRT.aspx), whereas the other computer controlled the eyetracker camera and collected the eye movement data. The experimenter viewed the infant’s eye movements and stimulus presentation on the data collection computer as a picture-in-picture video, via video capture software. The stimulus-generating computer sent a unique, time-stamped numerical code through a parallel port to the data-collecting computer, indicating the onset of a trial and the type of trial (e.g., cue with target and distractor in the spatial-cueing task, or an invariant color stimulus on either the right or the left side of the screen in the visual expectation task). Synchronization of the unique code with the eye movement data in the data file allowed coordination of the eye movement sequences with specific stimuli and their onsets.
Infants viewed stimuli in a sequence based on the Posner (1980) cueing paradigm. In the cued conditions, a trial began with the presentation of a fixation hexagon for 1,150 ms (see Fig. 2A). During the last 150 ms of the presentation of the fixation hexagon, a spatial cue (white triangle with a black outline) was presented at the location where the target would subsequently appear. Immediately after the presentation of the fixation and the cue, an interstimulus interval (ISI) of 2,500 ms was presented, during which the monitor was blank. A previous study by Gilmore and Johnson (1995) had shown that older infants exhibit eye movement facilitation to a cued target in the presence of a distractor with delays up to 3,000 ms. Furthermore, as was indicated by Johnson (2002), attentional shifts to the stimulus opposite the cued target due to inhibition of return is evident in infants 4 months of age and older, but likely not in younger infants. Considering that 3-month-olds are more likely to demonstrate sticky fixation than are older infants (Hunnius & Geuze, 2004), we chose this ISI so that infants would have sufficient time to disengage attention from fixation and still show facilitation to a cued target. Following the ISI, the target stimuli were presented, appearing on either the right or the left of the screen (target-only condition) or on both sides (target + distractor condition). The target stimuli were red or green Xs presented at a visual angle of 5.5° from the visual center of the screen. For the uncued conditions, the sequence of stimulus presentation was similar, but no spatial cue was presented and only the target-only condition was run. In all conditions, the latencies of the infants’ reactive eye movements to the target were measured.
Data reduction and analysis
The raw digital data recorded by the eyetracker were imported into a MATLAB toolbox called ILAB for analysis (Gitelman, 2002). ILAB allows for the analysis of eye movements, by parsing out and individually displaying the horizontal and vertical components of the eye movement data on a trial-by trial basis. The scan path of the eye for each trial was also displayed by ILAB, thereby allowing for the analysis of the nature of the eye movements (timing, direction, and distance) relative to the stimuli. With the use of ILAB, a scorer identified which of infants’ eye movements were anticipatory or reactive in timing.
In order for an eye movement to be included in the final data sample, it needed to meet a set of criteria. First, only the data of infants who attended (i.e., looked at the stimuli) on a minimum of 50 % of the experimental trials were included. Second, infants were required to be fixating the central stimulus during cue presentation and before onset of the target display. The purpose of requiring the infants to remain fixated was to allow for an assessment of the scan path of each eye movement from a single landmark that was a fixed distance from the target and, thus, to eliminate bias from producing eye movements that originated from random locations on the screen. Because infants cannot be told to remain fixated and can freely move their eyes at any time, this criterion ensured that the initial conditions for assessing infants’ performance were comparable across infants, conditions, and trials. Third, eye movements were considered to be anticipatory if they occurred after the offset of the previous stimulus and within the first 167 ms after the onset of the next stimulus. This latency value was designated as the anticipatory cutoff because previous studies had determined that 3-month-old infants cannot make reactive eye movements to the onset of a stimulus faster than 167 ms (Adler & Haith, 2003; Canfield, Smith, Brezsnyak, & Snow, 1997). If an eye movement occurred in the period from 167 ms after onset of the stimulus to 167 ms after stimulus offset, it was classified as a reactive eye movement. Finally, the eye movement to a stimulus had to trace a path that was more than 50 % of the distance to the intended stimulus from that eye movement’s starting location. This was assessed through the infant’s scan path in conjunction with the stimulus location. The 50 % criterion has been used in previous studies based on infants’ eye movements (Adler & Haith, 2003; Adler & Orprecio, 2006) and is typically taken as an indication that the eye movement is intentional and not random.
Due to the fact that one cannot require an infant to look on a given trial, the possibility existed that any given infant might not provide useful data in a particular condition. In order to account for potential missing data points and to increase the power of our statistical tests, the individual trial data from all the infants were pooled for each Experimental × Birth Experience condition, and analyses were based on the pooled data. This is consistent with a number of adult and infant studies that have used saccade latencies as the dependent measure (e.g., Adler et al., 2002; Adler & Gallego, 2014). On the basis of previous infant eye movement studies, to obtain a power (1 – β) equal to .90, a minimum of 21 observations per condition were required. With the pooling of the data, the smallest number of observations in any condition was 79, and the most was 106 (M = 95.0, SD = 10.1).
In typical spatial-cueing tasks, evidence of the role of bottom-up spatial attention in performance is provided by a comparison of responding to a target presented at a cued location relative to responding to a target presented at an uncued location (Klein & Lawrence, 2012; Theeuwes & Belopolsky, 2010). When spatial attention is facilitated by the cue, this comparison yields evidence that responding to a cued target location is speeded relative to an uncued location. With respect to eye movements, because they are intimately linked to spatial attention, the shifting of attention to the target location by the cue would speed eye movement initiation to the target (e.g., Adler et al., 2002). To assess the role of bottom-up attention in vaginally and caesarean-section-delivered infants’ eye movement responding, a 2 × 2 analysis of variance (ANOVA) was conducted on infants’ mean pooled latencies, with Cue Condition (cue–target-only and no-cue–target-only) and Birth Experience (vaginal delivery and caesarean-section delivery) as between-subjects factors. This analysis revealed that the main effect of cue condition was significant, F(1, 381) = 6.17, p < .02, Cohen’s f = 0.12, indicating that saccade latencies to a cued target (M = 745.82 ms, 95 % CI [688.61, 803.02]) were significantly faster than saccade latencies to an uncued target (M = 845.63 ms, 95 % CI [781.72, 909.53]). This result demonstrates, consistent with previous spatial-cueing studies (Klein & Lawrence, 2012; Theeuwes & Belopolsky, 2010), that the cue facilitated the allocation of spatial attention prior to presentation of the target, thereby speeding the eye movement response to the target.
The main effect of birth experience, F(1, 381) = 8.91, p < .01, Cohen’s f = 0.14, was also significant, indicating that the saccade latencies of infants delivered vaginally were faster (M = 762.76 ms, 95 % CI [677.72, 787.8]) than the saccade latencies of infants delivered by caesarean section (M = 855.67 ms, 95 % CI [789.61, 921.72]). As can be seen in Fig. 3, the interaction between cue condition and birth experience was not significant, F(1, 381) = 0.37, n.s., indicating that the saccade latencies of vaginally delivered infants relative to infants delivered by caesarean section were faster whether the target was cued (vaginal, M = 694.27 ms, 95 % CI [621.65, 766.89]; caesarean section, M = 796.88 ms, 95 % CI [708.51, 885.25]) or was not cued (vaginal, M = 775.31 ms, 95 % CI [691.19, 859.42]; caesarean section, M = 930.19, 95 % CI [833.64, 1,026.74]). These results demonstrated that, irrespective of whether or not spatial attention was cued, infants who were delivered by caesarean section took longer to allocate attention and initiate that eye movement than those who were delivered vaginally. Consequently, the lack of an interaction suggests that the significantly slower initiation of saccadic eye movements by infants delivered by caesarean section was not a consequence of a unique difficulty in processing the brief cue. Instead, the slower eye movements were likely due to differences between caesarean-section and vaginally delivered infants’ mechanisms for initially allocating spatial attention.
To further assess the effect of birth experience on infants’ bottom-up spatial attention, we analyzed infants’ latencies to initiate saccadic eye movements to a visual target’s location as specified by a prior spatial cue, when there was competition for attention relative when there was not. A 2 × 2 ANOVA was conducted on infants’ mean pooled latencies, with Target Condition (target-only and target + distractor) and Birth Experience (vaginal delivery and caesarean-section delivery) as between-subjects factors. This analysis revealed that the main effect of birth experience, F(1, 352) = 6.88, p < .01, Cohen’s f = 0.13, was significant, indicating that the saccade latencies of infants delivered vaginally were faster (M = 688.64 ms, 95 % CI [635.75, 741.53]) than the saccade latencies of infants delivered by caesarean section (M = 802.25 ms, 95 % CI [735.89, 868.6]). The main effect of target condition was not significant, F(1, 352) = 0.01, n.s., indicating that saccade latencies to a target presented alone (M = 744.48 ms, 95 % CI [683.88, 805.09]) did not differ significantly from saccade latencies to a target presented with a distractor (M = 746.47 ms, 95 % CI [685.97, 806.97]). As can be seen in Fig. 3, the interaction of target condition and birth experience was also not significant, F(1, 352) = 0.08, n.s., indicating that the differences in saccade latencies of vaginally delivered infants relative to infants delivered by caesarean section were similar when a target was presented alone (vaginal, M = 682.42 ms, 95 % CI [608.44, 756.4]; caesarean section, M = 807.91 ms, 95 % CI [711.78, 904.04]) and when a target was presented with a distractor (vaginal, M = 695.45 ms, 95 % CI [618.28, 772.61]; caesarean section, M = 794.53, 95 % CI [700.32, 888.75]). Thus, irrespective of whether infants had to initiate a saccadic eye movement to a single item or to a stimulus item in the presence of a competing item, those who were delivered by caesarean section took longer to allocate attention and initiate that eye movement than did those who were delivered vaginally.
Alternatively, the slowed attentional responses of caesarean-section infants relative to vaginal infants might have been due to other factors. Often, for example, when speeded responses are made, there is a speed–accuracy trade-off in which higher accuracy leads to slower responses, whereas faster responses lead to lower accuracy (Ho et al., 2012; Pachella, Smith, & Stanovich, 1978). Consequently, both groups of infants’ accuracies in making a saccade to the cued target in the target + distractor condition were computed. Contrary to a speed–accuracy trade-off explanation, caesarean-section-delivered infants made errors (24.18 %, 95 % CI [19.35, 29.00]) at essentially the same rate as vaginally delivered infants (29.45 %, 95 % CI [22.64, 36.26]), t(6) = 2.01, n.s., indicating that the caesarean-section infants were not slowing the initiation of their saccadic eye movements in order to more accurately select the cued target. Because of the relatively small sample size, additional data would be required to statistically confirm this finding.
An additional factor that could potentially influence differences between vaginally delivered and caesarean-section infants might be maternal age. Increased maternal age has previously been shown to be a factor in a number of adverse and atypical effects (Koyama, Kamio, Inada, & Inokuchi, 2011; Menezes et al., 2010; Sandin et al., 2012). Furthermore, increased maternal age is also correlated with an increased likelihood of delivering by caesarean section (Bayrampour & Heaman, 2010). Consequently, caesarean-section infants’ slowed attentional responding might have been due to influences of maternal age rather than to consequences of their birth experience. To assess this possibility, a 3 × 2 ANOVA was conducted comparing the mean maternal ages in each of three spatial-cue and no-cue conditions as a function of whether an infant’s birth experience was a caesarean section or a vaginal delivery. The results indicated no significant main effect of birth experience, F(1, 14) = 0.04, n.s., nor was the main effect of cueing condition significant, F(2, 14) = 0.45, n.s. These results indicate that maternal age was not different for the caesarean-section infants (M = 33.8 years, 95 % CI [30.00, 37.60]) than for vaginally delivered infants (M = 34.1 years, 95 % CI [30.81, 37.39]), nor was maternal age different for the different spatial-cueing conditions. The interaction between birth experience and cueing conditions was also not significant, F(2, 14) = 0.16, n.s., indicating that maternal age was the same for the caesarean-section and vaginally delivered infants in each cueing condition. These findings support the interpretation that caesarean-section infants’ slower initiation of saccadic eye movements relative to vaginally delivered infants was not due to complications of increased maternal age. Instead, considering that providing a cue speeded eye movements relative to when there was no cue (see Fig. 3) for infants delivered either by caesarean section or vaginally, this suggests that birth experience likely had an impact on the development of the brain area involved in bottom-up, stimulus-driven spatial attention. To reliably implicate bottom-up spatial attention mechanisms, however, would likely require the inclusion of additional conditions such as invalid cues. Furthermore, though the results of this experiment suggest a connection between birth experience and attention and eye movements, because birth experience is not a variable that can be manipulated, a causal relation with attentional responding cannot be assumed.