Probing Two Novel Proxy Markers of Excitation and Inhibition within the EEG Signal

Background: An optimum ratio of neural excitatory and inhibitory signals is necessary for coherent brain function. It has been proposed that autism is associated with an excitation: inhibition (E:I) imbalance. We investigated two novel candidate EEG markers of excitation and inhibition. Both measures tap into ‘neural noise’ so we expected variance in these markers to map onto adaptive functioning and intellectual disability. Inter-trial coherence (ITC) is a measure of neural consistency; higher values mean that brain activity is more temporally aligned across trials (Makeig et al., 2004). When inhibition is reduced, the brain may not be in the optimal state to process a relevant stimulus (Levin & Nelson., 2015) leading to lower ITC. The second proxy measure, 1/f, characterizes the steepness of the power spectral density (a plot of power at each frequency) distribution. Evidence from a computational model and preclinical experiments suggests that increased inhibition leads to steeper frequency decay (Gao, Peterson & Voytek., 2017).

Objectives: To carry out case-control analysis (ASD versus TD) to determine whether, on average, both metrics differ according to diagnosis. Given both measures tap into ‘neural noise’ we hypothesized that these measures would vary between individuals with and without intellectual disability, and correlate with adaptive functioning.

Methods: We analysed EEG data recorded during the auditory oddball task as part of the EU-AIMS LEAP consortium (age 6-30; 1/f analysis: 262 ASD and 195 TD participants; ITC analysis: 224 ASD and 182 TD). For ITC, the reduced sample size reflects removal participants who did not show an ERP (i.e. no phase locking); however, all results maintained when these participants were included. We also created a ‘level of functioning’ subgroup, by comparing individuals with IQ above (high-functioning; N=386) and below (low-functioning N=71) 75.

Results: There was no overall significant difference in ITC between ASD and TD groups (t= 1.484, p =0.128), and there was no difference in ITC between high-functioning and low-functioning groups (t = -1.069, p= 0.286). For children, ITC was significantly higher in the TD child group (M= 0.196) than the ASD child group (M= 0.164), (t= 2.016, p= 0.046, d=0.399). Across all participants, higher ITC scores were associated with higher adaptive functioning scores on the VABS (r= 0.354, p= <0.001; Figure 1). For 1/f, there was no significant difference between the ASD and TD group (t=1.280, p= 0.202), but there was a significant difference between the high-functioning and low-functioning groups (t = 1.010, p= 0.002, d= 0.403). Whereby, irrespective of diagnosis, the slope was flatter for the low-functioning (M= -1.165, SD= 0.406) than the high-functioning group (M= -1.317, SD= 0.345). 1/f correlated with adaptive functioning (r=-.18, p=0.002). For both the ASD and TD groups, the histograms for both 1/f and ITC were overlapping.

Conclusions: Neither ITC or 1/f slopes differed at the case-control level between autistic and non-autistic individuals. However, ITC values were associated with adaptive functioning, and 1/f slopes were on average flatter in the group with intellectual disabilities. Thus, both markers may reflect adaptive skills, but not a clinical diagnosis of autism.