Methods: Structural MRI volumes came from individual infants and from average MRI templates (3, 4.5, 6, 7.5, 9, and 12 months). The MRI volumes were either 3.0T 1x1x1 3D MRIs done at the McCausland Center for Brain Imaging (MCBI), or 1.5T 1x1x3 2D MRIs from the NIH longitudinal study of MRI (NIHPD).

First, a database of scalp locations to cortical locations was made. The 'Virtual 10-10' electrode locations were done on the scalp of individual infant or average MRI head volumes. These locations were projected inward to the extracted cortex of the MRI (Figure 1). Stereotaxic atlas volumes were used to determine the anatomical location of the projected 10-10 optode. 'Lookup tables' were constructed that have anatomical locations separated by each 10- 10 location, age, average or aggregated individuals, for four stereotaxic atlases. There was mild consistency across the averages and aggregations for the identification of cortical anatomical locations projected to by each 10-10 optode location.

Second, methods were developed for identifying optode locations and cortical projections for individual participants. This was done for participants for whom both a structural MRI and a NIRS recording was done (Lloyd-Fox et al, 2013).

Results:
The NIRS holder / optode / channels were located on the scalp for each participant MRI and the underlying cortical anatomy was identified. The identification of optodes was done for a group of infants that had a NIRS recording but did not have a structural MRI (Emberson et al, 2013, submitted). In this case the optode locations on the individuals were used to place optode locations on an age-appropriate average MRI template, or on an individual MRI with a head size similar to the NIRS participant. Figure 2 shows this NIRS holder on a 3-d rendered scalp, and the corresponding channel projections on the brain. There were discrepancies between the cortical anatomical locations based on the average MRI and individual MRI; suggesting that individual structural MRIs give some advantage to cortical identification.

Finally, these methods may be used for participants for whom the 3-D spatial locations are known, by registering the spatial locations of the optodes to the participant's own MRI, to an age-appropriate average MRI, or a MRI from a 'library' of infant MRIs.

Methods: Across both ongoing studies, 54 infants 12 months of age (17 FXS, 17 SIBS, 20 TD) participated. Autism symptoms were assessed using the Autism Observation Scale for Infants, and ASD diagnoses were determined using the Autism Diagnostic Observation Schedule 2. Event related potentials were measured in response to images of the child’s own mother (familiar social),favorite toy (familiar object) contrasted to an unfamiliar woman, and unfamiliar toy (500 ms duration). High-density EEG’s were recorded using a 128-channel net. Grand averages and peak amplitudes were calculated to capture the N290. The Nc was calculated from 350 to 700 ms post-stimulus onset. Heart activity was also collected during a passive viewing task as part of the larger battery in a subset of participants (9 FXS, 15 SIBS, 12 TD). Dependent measures included inter-beat interval, respiratory sinus arrhythmia, and proportion of time in heart-defined sustained attention, a period of decelerative heart rate that is associated with greater stimulus engagement in typically developing infants (e.g.Richards & Casey, 1991). We anticipate increased samples for both studies for the presentation.

Results: Results from the ERP study indicated a main effect for trial (F(1,88)=4.96; p<.01) with a larger amplitude for faces than toys for all groups for the N290. Results also indicated a marginal interaction effect (F(2,33)=2.79; p<.08) suggesting a larger effect for the toy than face stimuli for the FXS and SIBS groups. Also, evidence suggested that infants with FXS and elevated ASD symptoms showed a larger Nc to the stranger than to their mother that was not observed in the TD, SIBS and the FXS group with a low number of ASD symptoms. Preliminary heart activity data indicated that groups did not differ in inter-beat interval, respiratory sinus arrhythmia, and qualities of heart-defined sustained attention (Kruskal-Wallis p>.10). However, across groups, higher AOSI scores were associated with greater proportion of time in heart-defined sustained attention (r=.36, p=.03) and increased behavioral looking time (r=.41, p=.02).

Discussion: The larger P400 in the SIBS and FXS groups may reflect an object-based preference for processing. Also, the heightened Nc response to the stranger in the FXS group with elevated ASD symptoms could signal a relationship of ASD to neural components associated with attention and cognitive processing. Heart activity data indicate that increased heart defined sustained attention predicted elevated ASD symptoms across groups. These preliminary results indicate that biomarkers are sensitive to detect early markers of risk for ASD in vulnerable populations.

Methods: Structural MRI volumes came from individual infants and from average MRI templates (3, 4.5, 6, 7.5, 9, and 12 months). The MRI volumes were either 3.0T 1x1x1 3D MRIs done at the McCausland Center for Brain Imaging (MCBI), or 1.5T 1x1x3 2D MRIs from the NIH longitudinal study of MRI (NIHPD).

First, a database of scalp locations to cortical locations was made. The 'Virtual 10-10' electrode locations were done on the scalp of individual infant or average MRI head volumes. These locations were projected inward to the extracted cortex of the MRI (Figure 1). Stereotaxic atlas volumes were used to determine the anatomical location of the projected 10-10 optode. 'Lookup tables' were constructed that have anatomical locations separated by each 10- 10 location, age, average or aggregated individuals, for four stereotaxic atlases. There was mild consistency across the averages and aggregations for the identification of cortical anatomical locations projected to by each 10-10 optode location.

Second, methods were developed for identifying optode locations and cortical projections for individual participants. This was done for participants for whom both a structural MRI and a NIRS recording was done (Lloyd-Fox et al, 2013).

Results:
The NIRS holder / optode / channels were located on the scalp for each participant MRI and the underlying cortical anatomy was identified. The identification of optodes was done for a group of infants that had a NIRS recording but did not have a structural MRI (Emberson et al, 2013, submitted). In this case the optode locations on the individuals were used to place optode locations on an age-appropriate average MRI template, or on an individual MRI with a head size similar to the NIRS participant. Figure 2 shows this NIRS holder on a 3-d rendered scalp, and the corresponding channel projections on the brain. There were discrepancies between the cortical anatomical locations based on the average MRI and individual MRI; suggesting that individual structural MRIs give some advantage to cortical identification.

Finally, these methods may be used for participants for whom the 3-D spatial locations are known, by registering the spatial locations of the optodes to the participant's own MRI, to an age-appropriate average MRI, or a MRI from a 'library' of infant MRIs.

Introduction. Sustained attention is one type of the endogenous attentional function that affects infant cognitive processes. Faces are socially significant and ubiquitous in humans’ daily lives. Both Infant sustained attention and face perception develop dramatically in the period of 3 to 6 months. Electroencephalograph (EEG) and Event-related Potentials (ERPs) have been used as measurements of cortical responses in both infant attention and face perception studies (e.g., the Nc, N290, P400). Interstimulus Interval (ISI) is the time period between the offset of one stimulus presentation and the onset of the next stimulus presentation. The effect of ISIs on infant sustained attention is still unknown. The relationship between infant attention development and face perception development has not been examined either. This study examined the effects of various ISIs on infant sustained attention and face perception. Infant face perception development and its relation to infant attention development were also investigated.

Methods. Three types of ISIs were employed in this study. They were 400 -600 ms, 600 -1000 ms, and 1500 ms to 2000 ms. Three types of stimuli including female faces, infant faces, and objects will be used in this study. Infants at 3, 4.5, and 6 months of age will be recruited in the present study (N = 8). Electrocardiogram (ECG) and EEG data were recorded in this study. A high-density, 128-channel EEG recording system produced by the Electrical Geodesics Incorporated (EGI, Eugene, OR) was used for EEG recording. The The ERPs data were obtained from the EEG recordings, and three ERPs components are measured: Nc (attention), N290 and P400 (face processing).

Results. Infants have larger N290 and P400 responses to faces than toys (Figure 1). Figure 1 also indicates that three months show similar N290 responses to faces and toys but have larger P400 responses with shorter latency to faces than toys. By contrast, six months show clear larger N290 and P400 responses to faces than toys. Figure 2 shows that infants have different cortical responses to faces in different attentional phases defined by their heart rate data. Specifically, the N290 and P400 response to faces is smaller in amplitude and slower in latency when infants are paying attention. In the three types of ISIs conditions infant have different N290 responses to stimuli: N290 responses to stimuli (faces and toys) increase as the ISIs increase (Figure 2). In addition, faces and objects presented within the short ISI condition did not elicit clear different N290 and P400 responses for 3-month-olds; however, faces and toys in the short ISI condition did cause clear different ERP responses for 6-month-olds.

Discussion. Face-sensitive ERPs components (N290, P400) exhibit young infant distinct cortical responses to faces than objects. To our knowledge, this is the first study that possibly show larger P400 responses to faces than objects in 3- to 6-months infants. The findings of the N290 are consistent with previous research (e.g., Halit et al., 2003, 2004). Comparison of these ERPs between 3 and 6 months also indicates the development of infant face processing. Different patterns of ERPs responses in different attentional phases might suggest that infant attention has impact on their face processing. However, they might also due to unequal trials in different attentional phases. Another interesting finding is that short ISIs might be used in ERPs experiments with older infants; however, for young infants (e.g., 3 months), high rate presentations might not give them enough time to fully process target information.

Specialized processing of faces begins early in life, yet we are just beginning to understand the neural underpinnings of the development of face expertise in infancy. Adults exhibit differential neural responses to faces as opposed to other classes of objects, evidenced by a larger N170 amplitude for faces than for objects. The N170 is a negative event-related potential (ERP) component occurring about 170ms post-stimulus onset, and research with infants has found two components that may be precursors to the adult N170. One of these components, the N290, is a negative deflection over posterior regions peaking at 290ms that is greater in amplitude for faces than visual noise (Halit, Csibra, Volein, & Johnson, 2004). In the first year of life, as infants acquire extensive exposure to faces, the N290 begins to differ in response to upright (as opposed to inverted) faces and to human (as opposed to monkey) faces (de Haan, Pascalis, & Johnson, 2002). However, few studies have examined whether changes occur in the morphology of the N290 that correspond to emerging expertise with faces as opposed to other objects. In the current study, infants’ ERPs were recorded while infants of three different ages (4.5, 6, and 8 months) passively viewed faces and objects (toys).

Thirty-eight infants (14 4.5- month-olds, 12 6-month-olds, and 12 8-month-olds) participated in the study. Infants viewed a series of brief stimulus presentations (500ms) of images of their own mother, another infant’s mother, their own toy, or another infant’s toy randomly interspersed across trials. High-density EEGs were recorded using an EGI 128-channel Geodesic Sensor Net. The EEG data was analyzed for groups of electrodes over occipito-temporal regions (e.g., around T5, T6, O1, O2, Oz) based upon previous infant studies that have found the N290 to be most prominent over these areas (de Haan et al., 2002). For each participant, ERP grand averages were computed for the time of the target onset, and the peak amplitude was derived using individualized time windows to capture each subject’s N290.

The dependent measure was the mean amplitude of the N290 component. A main effect of age was found [F(2, 88)= 5.81; p < .005*], and the increase in the amplitude of the N290 from 4.5 to 7.5 months can be clearly seen in Figure 1. There was also a marginally significant effect of trial type [F(1,88)=3.26; p < .07], as the amplitude was larger for faces than for toys (but the interaction was not significant). Thus, the N290 amplitude was larger for faces than for toys (see Fig. 1), and increased in amplitude across the age groups.

The current study documents a developmental change in the face-sensitive N290 corresponding to a time period when infants are developing expertise with faces. These studies suggest that, like adults, infants demonstrate special processing of faces compared to other objects, and that the N290 is similar to the N170 in that its amplitude is greater to faces than other objects.

Infants of 14 or 20 weeks of age were tested in a spatial cueing procedure. A central stimulus was presented until the infant fixated it, a cue was presented in the periphery for 300 ms, and then a “target” was presented on the same side as the cue (“valid”), on the opposite side of the cue (“invalid”), or a target was presented without a prior cue (“neutral”), and at short- or long- stimulus onset asychrony. Heart rate deceleration and the return of heart rate to a prestimulus level were used to define periods of attentiveness and inattentiveness. ERP averages were computed around the time of the target onset. A neurodevelopmental MRI database and “Finite Element Model” (FEM) head models were used for a realistic head model for source analysis. About half of the participants had a structural MRI, and the MRI from the participant or from an age-appropriate average MRI template was used for the head model. The cortical sources of the P1 validity effect was examined with current density reconstruction methods using sLORETA. Several posterior brain areas have been hypothesized to control the P1 validity effect in adults (e.g., lateral occipital cortex, fusiform gyrus, inferior-posterior temporal cortex). These and surrounding control areas (central occipital pole, superior parietal cortex, superior-posterior temporal cortex) were used with a “region of interest” (ROI) analysis to quantify current density in these regions as a function of experimental variables.

ERP analyses showed an enhanced P1 on valid trials over invalid and neutral trials. This P1 component was enhanced if the infant was attentive, and primarily on short SOA trials. The enhanced P1 was larger for short SOAs, suggesting this was spatially-cued orienting rather than endogenous attention. The current density reconstruction analysis was used with the hypothesized posterior brain regions to quantify cortical current sources. The P1 facilitation occurred both in the in the lateral occipital cortex and the inferior temporal cortex. However, attention modulated the P1 validity effect for the current density only in the lateral occipital cortex. Surrounding areas and control areas did not show the current density as a function of these experimental variables. These results suggest that the lateral occipital cortex and the inferior temporal cortex are involved in the modulation of the P1 validity effect by attention. This study demonstrates the use of realistic head models for cortical source analysis of infant spatial attention.

Methods. Twenty-two infants (6 TD, 10 ASIB, 6 FXS) were tested at 12 months of age as part of an ongoing study. Infants viewed paired comparison trials of simultaneously-presented faces (mother, stranger) or toys (participant’s toy, unfamiliar toy). Preference scores were calculated as the proportion of looking time toward the unfamiliar stimulus. Infants also viewed brief presentations (500ms) of each stimulus for the purpose of measuring stimulus-onset event related potentials (ERPs) implicated in novelty (Nc) and face processing (N290, P400).

Results. Behavioral data were analyzed using repeated measure analyses of variance (Figure 1). Within face trials, preference toward the stranger differed by group, F(2, 19)=6.80, p=.001, with FXS and ASIB groups showing less stranger preference than controls. Preference toward unfamiliar toys also differed by group, F(2, 19)=12.04, p<.001, with the FXS group showing less novelty preference than both TD and ASIB groups. Together, these behavioral data suggest a face-specific reduction in novelty preference in ASIBs and a more general reduction in novelty preference to both faces and toys in FXS.

Visual inspection of ERP grand averages suggests atypical Nc patterns in both high risk groups (Figure 2). Both FXS and ASIB groups produced larger Nc components than controls, with the greatest amplitudes in the FXS group. All groups showed greater frontal Nc amplitude toward familiar versus unfamiliar toys. Although the control group produced a slightly greater central Nc amplitude toward the stranger versus mother, this pattern was reversed in high risk groups in which slightly greater responses were observed toward mothers versus strangers. The peaks of the Nc and N290 occurred later in response to faces in ASIBs, whereas peaks occurred later in response to objects for FXS (Nc) and TD (Nc and N290) groups. These findings may suggest a developmental lag in shifting preferential attention toward strangers versus mothers in ASIBs and FXS, as well as atypical Nc and N290 latencies associated with processing faces in ASIBs.

Discussion. Atypical face processing is well-documented in autism and may be related to the socio-communicative deficits inherent to the disorder. Our data indicate both shared patterns of atypical face and object processing across ASIBs and FXS, as well as several face-specific patterns unique to ASIBs. Studying early markers of atypical visual processing across multiple high-risk groups may inform the latent heterogeneity of indicators, potentially contributing to early detection and intervention of efforts

The ages came from the neurodevelopmental MRI average brain templates (3, 4.5, 6, 7.5, 9, 12, 15, 18, 24, 30, 36, 48 months). A stereotaxic atlas distributed with the FSL computer program, the MNI atlas, was used. This atlas identifies the major cortical lobes, cerebellum, and some subcortical regions. The atlas was modified to fit an average brain template based on 20-24 year olds with high-resolution 3T scans (T1W, 3T, 3D, 1x1x1mm resolution). This was done by affine registering the MNI average brain template to the 20-24 year old average brain template, then warping the atlas from the MNI to the 20-24 year old and manually modifying the atlas to fit the average template. The 20-24 year brain template was registered to infant age brain templates, and the atlas was warped to the brain templates for ages 3, 4.5, 6, 7.5, 9, and 12 months (templates based on T1W, 3D, 1x1x1mm resolution, N = 10 at each age). The transformed atlas was manually edited to fit precisely on the average template for each age. The resulting atlas has the cortical lobes (frontal, parietal, occipital, temporal, insula), two gyri (fusiform, cingulate), cerebellum, brainstem, thalamus, striatum, corpus callosum, and ventricles. Atlases were constructed with a “majority fit procedure” from 3 month atlas to 2 week template, and from 12 month atlas to 15, 18, 24, 30, 36, and 48 months.

The atlases were tested by examining 4 individuals at each age with five manually identified areas (frontal, occipital, temporal, cerebellum, thalamus). The manually drawn areas were compared to registered/warped areas for the age-appropriate atlas, atlases from older infant ages, and adult atlases. There was a close fit between the manually identified and automatically labeled regions for the age-appropriate atlas (e.g., 3 mo individuals with 3 mo avg, DICE coefficient > .95; 12 mo individuals with 12 mo average, DICE coefficient > .97). The fit systematically decreased for older age atlases, and there was an expected poor fit with adult atlases (e.g., 3 or 6 mo with 20-24 year average, ~.85).

These atlases allow the automatic identification of anatomical areas in infant MRI studies. They may be used to guide the development of more detailed stereotaxic atlases (e.g., “Harvard-Oxford” gryal and sulcal identification; “AAL” stereotaxic atlas; LONI PBA40 atlas).