Prior studies have shown that performance on standardized measures of memory in children with autism spectrum disorders (ASD) is substantially reduced in comparison to matched typically developing controls (TDC) [
1,
2,
3,
4,
5,
6]. This is not surprising given speculation of various white matter temporal lobe abnormalities in ASD [
7] and the role that temporal structures play in memory and learning [
8]. One measure that has demonstrated multiple differences in memory functioning in those with ASD compared to TDC [
5,
9] is the Test of Memory and Learning (TOMAL) [
10]. The TOMAL is comprised of various verbal and non-verbal subtests, including the Facial Memory subtest. Performance on the TOMAL Facial Memory subtests may be of particular interest in studying memory impairments in ASD because of associated deficits in face processing [
11] and atypicalities in the fusiform gyrus (For simplicity, throughout the paper we will often just refer to the fusiform and not fusiform gyrus, gyral, or volume.) of the temporal lobe [
12]. Despite the large amount of research pertaining to facial processing in ASD, the literature examining facial memory is more limited, and a comprehensive understanding of facial memory functioning in this population is lacking [
13,
14,
15].
1.1. Facial Processing and Memory as a Deficit in Autism
Facial processing has been recognized as a specific deficit in autism [
16,
17,
18]. Individuals with ASD exhibit impairments in perception of facial affect [
19,
20,
21], direction of eye gaze [
22], eye contact [
23,
24], and attention to eyes [
25,
26]. Specific behavioral impairments in ASD help provide clues to the mechanisms of abnormal facial perception. For example, individuals with ASD tend to fixate longer on objects than faces (similar to typically developing individuals), but are less likely to scan regions of the face outside the primary facial features (
i.e., eyes, nose, mouth) [
27,
28,
29,
30]. Research also suggests that individuals with ASD may rely on individual features during facial processing rather than taking a holistic approach [
31] and also may tend not to benefit from face orientation during facial recognition tasks [
20,
32]. These results provide evidence for atypicalities in how individuals with ASD attend to and process facial stimuli.
Individuals with ASD have also been found to perform more poorly on facial recognition tasks relative to object recognition [
13,
14,
33,
34,
35]. This has been suggested to be due to abnormal scanning of facial regions during encoding, which certainly could be the basis for impaired facial memory [
29]. In a review of studies of facial perception in ASD, Weigelt
et al. [
36] reported that quantitative deficits in facial perception on behavioral tasks are much more impaired in tasks with memory demands, although other hypotheses have been proposed as well. For example, impaired facial perception may also be due to greater visuospatial effort required for facial processing [
37], the inherent social content of face stimuli [
38], or impaired gaze fixation [
25,
39]. Indeed, research on facial processing in ASD has revealed that individuals with ASD have impaired prototype formation of faces [
11], which may help explain why once faces are attended to, they may be categorized and consolidated in memory very distinctly in ASD.
Specific to facial memory, children with ASD are particularly impaired in their memory for faces [
15], which may be less apparent in adolescence [
14] and least apparent in adulthood [
14,
15]. Also, Kuusikko-Gauffin and colleagues [
35] found that facial memory improved with age, with significant differences in facial memory between children, but not adolescents or adults. In addition, parents of ASD individuals had poorer facial memory performance than control parents.
1.2. Neuroanatomical Correlates of Facial Processing and Memory
Functional neuroimaging studies have identified a face-specific region in the fusiform gyrus of the temporal lobe termed the fusiform face area (FFA) [
40]. The FFA is responsible for processing both facial features (e.g., nose, mouth, eyes), as well as the spatial relation among face parts [
41,
42,
43,
44,
45,
46,
47]. Disruption of the fusiform face area in the fusiform gyrus may help explain why individuals with ASD have deficits in facial processing and facial memory.
Kleinhans
et al. [
48], Anderson
et al. [
49], and Khan
et al. [
50] have shown reduced functional connectivity in ASD not only between the fusiform and other cortical areas but also between left and right fusiform gyri, and
within the fusiform gyrus itself during face processing. Studies that have examined individuals with identifiable lesions to the fusiform have also demonstrated similar facial processing impairments [
51,
52,
53]. Generally, the literature supports abnormalities in face-processing networks involving the fusiform, including reduced long-range and local functional connectivity (
i.e., within the fusiform face area) [
50], rather than only a region specific abnormality [
39,
54,
55].
The fusiform is also functionally related to both the amygdala and the hippocampus, two structures critical for memory and emotional processing. Studies looking at face processing, rather than memory, have found abnormal pathway microstructure [
56] and connectively [
50] between the hippocampus/amygdala and fusiform. All three of these regions show reduced activation during task-based fMRI studies of facial processing in autism [
25,
57,
58,
59]. Hypoactivation of the amygdala and fusiform is often observed in ASD individuals relative to TDC [
55]. It is also important to consider that the lack of activation in these brain regions in individuals with ASD compared to TDC may also relate to differences in processing emotional intensity [
60] and dynamic
versus static facial stimuli [
54,
61]. Both facets of facial processing (
i.e., emotional intensity and dynamic expressions) have more ecological validity pertaining to social interaction deficits than the simple viewing of pictures of facial expressions (
i.e., static stimuli). Thus, a multitude of factors may influence face processing and facial memory.
Several studies have examined fusiform gyral volume comparing controls to those with ASD. Volume in TDC individuals is considered a marker of structural integrity, albeit a coarse indicator of brain development [
62]. ASD studies that have examined fusiform gyral volume have reported differences in size [
63,
64,
65,
66,
67,
68,
69,
70]. However, the direction of the differences, including hemispheric effects, varies. In a meta-analysis, Cauda
et al. [
71] reported a larger fusiform associated with autism, but underscored the variability of reported differences across studies. Inconsistencies in the volumetric literature in autism also exist for other temporal lobe structures such as the amygdala and hippocampus. In a sample of individuals with Asperger syndrome, Murphy
et al. [
72] found larger amygdala but not hippocampal volume. In contrast, Hasan, Walimuni and Frye [
73] reported larger hippocampal volume in autism. The lack of a consistent direction to volume differences in the fusiform may be a reflection of the heterogeneity of ASD and associated morphological differences that may also result in varied cognitive impairments.
Additionally, variability in volumetric findings of temporal lobe structures in ASD likely has to do with age, developmental, and maturation effects. For example, some volumetric studies have implicated early overgrowth followed by normalization of amygdala volumes in middle childhood with either normalization or persistence of hippocampal enlargements [
56]. Some studies only examined adults, like Dziobek, Bahnemann, Convit, and Heekeren [
68], who found that the relationship between amygdala volume and fusiform thickness was actually smaller in autism compared with TDC, providing further evidence for disrupted neural networks. Still others have found reduced volume of the hippocampal–amygdala complex in autism in adolescents and adults [
63,
74]. As such, variability in reported volumetric findings of temporal lobe structures in ASD likely reflects differences in age and heterogeneity of the disorder. Pelphrey, Shultz, Hudac, and Vander Wyk [
55] propose that “ASD begins with a failure in the emergence of the specialized functions of one or more of the set of neuroanatomical structures involved in social information processing. This failure happens early in ontogeny, within the first nine months to one year of life, if not earlier. In turn, because the affected regions do not generate the normal stream of both intrinsic and stimulus driven signals, the normal developmental pattern of connections among these brain regions is greatly altered” (p. 4). Thus, examining volumetrics, although not a direct measure of neural connectivity, represents a logical place to begin in understanding how it might influence abnormal functioning of specialized neuroanatomical structures (and subsequently connectivity and functionality).
How fusiform morphology may contribute to impairments in facial memory is not yet known. However, it would seem to be a logical structure for examination, given the role the fusiform plays in face processing. Likewise, because of the important role that the medial temporal lobe plays in memory—particularly the hippocampus and, to a certain extent, the amygdala—it would be important for any facial memory study to volumetrically assess these regions, as well. Accordingly, the current study investigated whether hippocampal, amygdala, or fusiform gyral volume related to performance on the TOMAL Facial Memory task, both immediate and 30-minute delayed recall, in children 5 to 19 years of age with ASD compared to TDC age-matched individuals. The format for assessing TOMAL Facial Memory includes an immediate recognition recall trial where previously observed faces have to be identified amidst foils not seen. With each trial, the number of target faces and foils increases. Because performance on this initial trial requires face processing, individuals with ASD would be expected to perform more poorly, possibly just because of the challenges specific to processing facial information.
The delayed recognition trial occurs 30 min after the immediate recall trial and is composed of faces that have been previously viewed along with foils that have not. The child has no opportunity for rehearsal during the 30-min interval. Since the previously seen face has already been initially processed, this delayed aspect of the TOMAL Facial Memory Task taps consolidation. As a contrast to facial memory, the TOMAL also utilizes visual memory tasks such as Visual Selective Reminding that has no aspect of face processing but rather visual spatial retention, using both immediate and delayed recall. Also, the Object Recall task assesses immediate retention of visually presented line drawings of common objects including a single drawing of a generic face as one of 24 stimuli. Object recall does not have a delayed retention measure. By comparing ASD and TDC participants on the TOMAL Facial Memory subtest with visually processed memory tasks like the Visual Selective Reminding and Object Recall provides the comparison of how specific an impairment in facial memory may be or whether more general non-verbal, visual memory impairments may be associated with ASD. Also, of importance is whether these TOMAL memory measures relate to fusiform, hippocampal and amygdala volume.
We examined several hypotheses about the role of fusiform gyral, hippocampal, and amygdala volume in TOMAL Facial Memory performance. First, it was hypothesized that facial memory performance would be significantly lower for individuals with autism than for TDC participants. Second, given the literature supporting volume differences in these temporal lobe structures in ASD, it was hypothesized that the fusiform gyral, amygdala, and hippocampal volumes would be larger for the ASD group when compared with TDC participants and, furthermore, that these structures would be negatively correlated with TOMAL Facial Memory performance, both immediate and with a 30-min delay; however, that fusiform gyral volume would not correlate with TOMAL Performance on the Visual Selective Reminding and Object Recognition.