Autism sensory dysfunction in an evolutionarily conserved system

There is increasing evidence for a strong genetic basis for autism, with many genetic models being developed in an attempt to replicate autistic symptoms in animals. However, current animal behaviour paradigms rarely match the social and cognitive behaviours exhibited by autistic individuals. Here, we instead assay another functional domain—sensory processing—known to be affected in autism to test a novel genetic autism model in Drosophila melanogaster. We show similar visual response alterations and a similar development trajectory in Nhe3 mutant flies (total n = 72) and in autistic human participants (total n = 154). We report a dissociation between first- and second-order electrophysiological visual responses to steady-state stimulation in adult mutant fruit flies that is strikingly similar to the response pattern in human adults with ASD as well as that of a large sample of neurotypical individuals with high numbers of autistic traits. We explain this as a genetically driven, selective signalling alteration in transient visual dynamics. In contrast to adults, autistic children show a decrease in the first-order response that is matched by the fruit fly model, suggesting that a compensatory change in processing occurs during development. Our results provide the first animal model of autism comprising a differential developmental phenotype in visual processing.


Introduction
Autism spectrum disorder (ASD) has a strong (albeit complex) genetic basis with a large number of genes implicated [1][2][3][4][5]. A variety of genetic animal models have been proposed for ASD, including murine models [6][7][8] and, more recently, fly models [9]. However, for an animal model of any disorder/disease to be useful it needs to fulfil as much face validity as possible (i.e. exhibit a similar phenotype to humans with the disorder/disease). This poses a challenge for multifaceted, heterogenic disorders having symptoms that are difficult to operationalize and measure in animals. While there have been some attempts at measuring defining behaviours of ASD in animal models [10], including difficult to assess social interactions [11], repetitive behaviours [12] and confined interests [13], the links between human symptoms and equivalent animal behaviours are tenuous. For example, social symptoms in mice have been evaluated as defensive behaviour against intruders [11], or as courtship call frequency and wing extension in fruit flies [9], even though neither behaviour manifests in humans.
In addition to the defining social and behavioural features of ASD, autistic individuals report a host of sensory symptoms including unusual sensory interests as well as hyper-and hyposensitivity to intense stimuli such as bright lights or loud noises [14,15]. These human ASD sensory-processing symptoms have been well-documented behaviourally [16 -18], with electroencephalography (EEG) [16,17] and neuroimaging [19], and can also be measured in animals using equivalent methods [20]. Functioning in sensory systems may be better conserved over evolution than more complex behaviours associated with ASD, therefore we pursued a comparison of sensory responses in humans with ASD and an Nhe3 fruit fly model of ASD.
A previous study in mice measured visual responses in a related developmental condition, Rett syndrome, and was able to link decreases in visual neural responses and poor visual acuity across species [21]. However, it is difficult to generalize these findings to ASD, as human Rett syndrome lacks the pervasive sensory symptoms characteristic of autism [22]. An advantageous alternative to rodent models is Drosophila, given the ease in developing genetic mutations and ability to test many individual animals. Successful Drosophila models of human neurological disorders have so far been developed for Parkinson's disease [23], fragile X syndrome [24] and Alzheimer's disease [25]. Fruit flies share 75% of human disease-causing genes [26] and have a visual system exhibiting similar nonlinear neural properties, including a colour-and luminance-selective module as well as a motion-selective module [27]. The neural dynamics of these modules closely resemble those of transient and sustained neural populations in humans [20,28,29]. These factors combine to provide an excellent framework for modelling changes in early sensory neuronal signalling [29] which may lie behind atypical sensory processing in autism.
In this study, we evaluated a genetic Drosophila model of human ASD by measuring comparable visual responses both in autistic humans and in mutant Drosophila. In humans, lossof-function mutations in the gene SLC9A9 have been linked to ASD [30]. Here, we used a Drosophila orthologue of SLC9A9 -Nhe3. A homozygous P-element insertion loss-offunction mutants (Nhe3 KG08307 ) and Nhe3 hemizygotes (Nhe3 KG08307 /Df(2 L)BSC187) were used to inhibit Nhe3 function in fly. The use of two Nhe3 mutations in different genetic backgrounds ruled out the possibility of other mutations influencing the flies' visual responses. To assess the functionality of the visual system in these species, we measured steady-state visually evoked potentials (ssVEPs) to temporally modulated contrast stimuli. During this paradigm, a stimulus flickered on/off at a particular frequency (for example, 12 Hz) while neural responses were recorded from the organism. Using Fourier transformation, we then converted time course data into the frequency domain where the amplitude of different frequency components of the neural responses can be measured. From there, we extracted the first harmonic response (which follows the stimulation frequency of 12 Hz), as well a second harmonic response. Second harmonics are responses generated by the brightening and darkening transients of the stimulus flicker, which is 24 Hz in the flies and to contrast onset/contrast offset in human. The first and second harmonics probe different aspects of the dynamics of the visual system: sustained and transient neural responses, respectively [31]. Previous genetic dissection of the fruit fly has localized the first harmonic to photoreceptors and the second harmonic to the lamina [20].
As the visual systems of humans and fruit flies are difficult to compare anatomically, the visual responses obtained here were produced by functionally equivalent human and fruit fly neural substrates. In each organism, we assessed the same functional mechanism-contrast transduction. This computation in the fly is performed at the level of photoreceptors and lamina, whereas in humans the same computation is performed in the retina and in the early visual cortex (V1). A similar cross-species computational equivalence in the face of vastly different neural substrates has been shown previously for motion perception: thirdorder correlations required for motion perception were found in the lamina of the fly and areas V1 and MT in humans [29].
Furthermore, to investigate the progression of ASD sensory atypicalities over the course of development, we also measured visual responses at two stages of fruit fly maturation and acquired similar responses from autistic children and adults. Finally, as the ASD phenotype is complex and non-binary, we validated our sensory model with a large sample of neurotypical participants with high and low numbers of autistic traits. To quantify this functional dissociation while controlling for overall responsiveness of the visual system, we calculated a normalized ratio between first (1F) and second (2F) harmonics ð1F À 2FÞ=ð1F þ 2FÞ and averaged over the highest contrast conditions (where the response rises above the noise floor; see Methods). This allowed us to measure the differences between sustained and transient responses while normalizing for overall responsiveness of the visual pathway. The ratio was significantly higher in both mutant strains than in the controls (ANOVA, F 2,33 ¼ 20.53, p , 0.0001, both paired contrasts p , 0.001; figure 2c). These data suggest an impairment in the post-receptoral neural structures (downstream of the photoreceptors) of the older mutant flies [32].

Results
Interestingly, unlike the older flies, the young 3-day-old flies showed a reduced response at both frequencies (figure 3a,b) relative to controls. Importantly, there was no effect of genotype on the ratio between harmonics (F 2,33 ¼ 1.  were similar at the first harmonic (a one-way ANOVA showed no effect of group F 2,33 ¼ 0.05, p ¼ 0.95) but (b) responses were reduced for P/P (simple contrast, p ¼ 0.025) and P/Df mutants compared to controls at the second harmonic (simple contrast p ¼ 0.001; ANOVA group effect F 2,33 ¼ 6.71, p , 0.01). (c) Ratios between frequencies ð1F À 2F=1F þ 2FÞ were significantly higher for P/P ( p , 0.001) and for P/Df ( p , 0.0001) than for the control genotype. First harmonic responses were also similar for the (d) high AQ and low AQ groups and (g) for autistic and neurotypical adults. However, second harmonic responses were reduced for both (e) adults with high AQ and (h) autistic adults compared to controls. (f,i) The ratio between harmonics was also higher in both experimental groups compared to controls ( p ¼ 0.005 and p ¼ 0.04, respectively). Curved lines are hyperbolic function fits to the data. Frequency ratios are baselined in respect to the mean over groups of each comparison for display purposes. Error bars in all panels represent + s.e.m.
using the Autism Spectrum Quotient (AQ) questionnaire [33] were tested with the ssVEP paradigm. Visual responses were recorded from an occipital electrode (Oz, located at the back of the head over the visual cortex) to grating stimuli flickered at 7 Hz. Seven contrast conditions (each repeated eight times) were presented in a randomized order. First and second harmonic ssVEP responses were again derived via Fourier analysis. The evoked response data were averaged separately over participants split by their median (median ¼ 14) AQ score: high (n ¼ 53, AQ mean ¼ 20.57, s.d. ¼ 6.66) and low (n ¼ 47, AQ mean ¼ 9.47, s.d. ¼ 3.08) AQ (high AQ implying many autistic traits). The second harmonic was notably reduced in the high AQ group, similarly to mutant fruit flies (figure 2d,e). In addition, the first harmonic response was slightly increased in the high AQ group. A two-way ANOVA showed the interaction between group and frequency to be significant (F 1,98 ¼ 6.17, p ¼ 0.015). The high AQ group also had a significantly higher frequency ratio than the low AQ group (t 98 ¼ 2.86, p , 0.01; figure 2f ). Moreover, a regression analysis showed that AQ scores correlated with the frequency ratio, with high AQ scores being predictive of higher ratios (R ¼ 0.26, F 1,98 ¼ 6.87, p ¼ 0.01; figure 4). This result shows a relationship between the amplitude of the second harmonic response and the severity of the subclinical ASD phenotype; however, this effect cannot be directly generalized to clinical autism as the AQ is not diagnostic of full-blown ASD.
(c) Adult autistic individuals show a similar pattern of responses as mature Nhe3 flies We assessed the ssVEP difference between harmonics in clinical ASD by testing 12 typical-IQ autistic adults whose diagnosis was confirmed with the Autism Diagnostic Observation Schedule, second edition (ADOS-2) [34], and 12 age-and gender-matched controls using the same human ssVEP paradigm. The pattern of data again mimicked that of the previous adult dataset: there was a significant interaction between group and frequency (   Considering the striking similarity between the adult human datasets and the adult fruit fly model, it is reasonable to ask if similarities also exist between human children and young ASD-mimic flies. Specifically, our fly model predicts that the visual system of autistic children should show reduced responses at both the first and second harmonics. To examine this, we recorded from 13 autistic children (5 -13 years old) and 17 neurotypical age-and gender-ratio-matched controls using an ssVEP contrast-sweep paradigm. Artefact rejection was employed to control for movement and blinking in both groups. The stimulus in each sweep trial increased continuously in contrast from 0% to 50% in logarithmic steps. Data were binned into nine contrast levels before being Fourier transformed to compute response amplitudes.
As predicted by the model, the ASD group showed reduced amplitudes of the 1F, sustained response (t 28 ¼ 2.07, p ¼ 0.04; figure 3d,e) which was not found in the autistic adults, individuals with high AQ or older mutant fruit flies. A two-way ANOVA also revealed a significant group effect over both frequencies (F 1,28 ¼ 4.23, p ¼ 0.049). Unlike adults, children exhibited no difference in frequency ratios between the groups (t ¼ 1.41, p ¼ 0.17; figure 3f ). Although children showed reduced amplitudes in the sustained response as predicted by the Drosophila model, the amplitude reduction observed in the fruit fly second harmonic responses was in the same direction, but was not statistically reliable in the children (t 28 ¼ 1.26, p ¼ 0.219). This may be due to difficulty in measuring the relatively smaller 2F response in children.

Discussion
We found sensory-processing alterations in our Drosophila model of ASD that were consistent with similar response alterations in human data at two stages of development. Our steady-state electrophysiology data showed a selective depression in second harmonic visual responses in autistic adults, individuals with high levels of autistic traits and Nhe3 mutant fruit flies, suggesting that this response alteration is specific to the autistic phenotype in mature individuals of both species. These differences were also present when we calculated first/second harmonic ratios in order to control for changes in overall visual sensitivity. This suggests that the transient component of visual processing is selectively affected. Autistic children and young Nhe3 flies showed an alteration in sustained visual processing, not present in the adults. The Nhe3 fruit fly model of autism was predictive of these sustained visual response alterations both in children and in adults (atypical in early life, normal in later life), suggesting a fundamental and pervasive change in visual processing occurs during development in ASD. Although the human Nhe9 is only one gene implicated in ASD, its orthologue in fruit flies was able to produce a measureable sensory processing effect, which has a close counterpart in human ASD.
We replicated the response alterations of autistic adults in neurotypical individuals with high AQ: this group had visual responses consistent with those of participants diagnosed with ASD, suggesting common visual response properties between samples. This was unsurprising as previous research has found that AQ scores in the general population are highly correlated (R ¼ 0.77) with sensory processing difficulties, as measured by the Glasgow Sensory Questionnaire [35], indicating that high-AQ individuals exhibit milder forms of sensory difficulties.
The intact first harmonic response in adult flies and humans indicates normal functioning of mechanisms which give rise to the sustained response. Conversely, the reduced second harmonic response, as well as the increased ratio between harmonics, suggest a modification in the transient dynamics of the visual system. In flies, the first harmonic has been associated with sustained photoreceptor polarization and the second harmonic with second-order lamina cells [20]. In humans, an association has been made between simple cell and sustained responses to pattern onset and between complex cells and transient responses at both stimulus onset and offset [36]. Although simple cells exhibit some transient response properties as well [36,37], the intact first harmonic of adults suggests that their response modification is specific to human complex cells that only generate evenorder response components. This early, cell-type-specific deficit may explain previous findings of atypical neural dynamics of spatial frequency processing in ASD in the face of normal sensitivity thresholds [38,39].
Mechanistically, lower second harmonic responses could either be generated by disturbances in nonlinear transduction of visual signals or by subsequent temporal processing. As the second harmonic, by definition, has a higher temporal frequency, a bandpass temporal filter shifted towards lower frequencies would attenuate signals at this frequency more compared to the first harmonic. There is at present no consistent evidence for lowered temporal resolution/prolonged integration in human ASD. One study found no difference  between autistic and neurotypical participants [40], one study found finer/higher temporal resolution [41] and another found coarser/lower temporal resolution [42]. The possible role of temporal integration time could be tested in future work by using a lower stimulus frequency (such that the second harmonic would now equal the current first harmonic frequency) and by observing whether the difference between harmonics disappears. An absence of a difference would indicate that temporal filtering is affected in ASD, whereas a persistently reduced second harmonic would indicate a difference in the nonlinearity. The differences in sustained and transient modules observed in our Nhe3 model mimics the alteration of neural dynamics in autistic adults. Nhe3 affects the exchange of sodium and hydrogen ions in cell membranes directly affecting neural signalling [30,43]. Differential expression of Nhe3 and other genes in ASD, which has been observed in other parts of the brain [30,44], may extend to differential expression in colour and motion modules in the Drosophila visual system. As Nhe3 (SLC9A9 in humans) is only a single gene in a multifaceted genetic etiology of autism, it is likely that the expression of several genes in human autism affects simple and complex cell dynamics, producing similar effects at the neural population level. Furthermore, such abnormality in gene expression in other parts of autistic brains, as well as environmental influences and gene -environment interactions, may give rise to a wide range of cognitive and social differences in childhood and adulthood.
Our data indicate little or no over-responsivity in the visual responses that are predicted by excitation/inhibition (E/I) imbalance theories [45,46] and consistent with measurements of some previous studies [47,48]. However, it is possible that an E/I imbalance in autism stemming from GABA-ergic mechanism differences affects different neuron types or processing pathways in distinct ways and to different extents. It is also possible that E/I imbalance in sensory cortical areas in autistic individuals compensates for lower sensory signals (such as the second harmonic response here) in childhood. Regardless, cell-type-based processing modifications may explain previous inconsistencies in studies of sensory symptoms in ASD that did not differentiate the relevant neural dynamics [17]. Furthermore, the current results can provide an amended explanation to the magnocellular (M pathway) dysfunction hypothesis [16]. As it is difficult to isolate the M pathway by changing stimulus properties [31], the paradigms previously used to investigate magnocellular dysfunction in ASD may have been selectively activating responses of transient components rather than the M pathway, in particular [16,49].
Developmentally, the observed lessening of the response modifications in both species with increasing age is in accordance with previous findings showing reduction or complete rescue of neuroanatomical differences present in early ASD childhood over the course of maturation [50]. Previous longitudinal research has also shown that symptom severity in individuals diagnosed with ASD in childhood decreases over time [51,52]. McGovern & Sigman [52] found that 48 adolescents, who were diagnosed with ASD as children, showed marked improvement in social interaction, repetitive/stereotyped behaviours and other symptoms, with two no longer meeting criteria for ASD under ADI-R criteria, and four under ADOS criteria. This might be explained by a change in neural processing during development, which would be likely to affect both complex behavioural and simpler sensory outcomes.
One possible mechanism that would explain the developmental change is that the atypical nature of neural signalling (such as ion balance in the case of Nhe3) changes over time. In flies, reduced Nhe3 expression may reduce the rate at which sodium ions and protons are exchanged across the cell membrane. At least in mosquitoes, this exchanger is found in the gut, and Malpighian tubules (the fly equivalent of the kidney) [53]. Failure to properly regulate ionic balance in young adult flies might affect the sodium concentration, or proton levels in the body and brain, and affect the speed and intensity of action potentials. Later in life, the normal balance may be restored. A similar reduction in efficacy of SLC9A9, linked to ASD, may also be present and explain the homology. In this respect, we note that another transporter, the potassium/chloride exchanger, has been linked to epilepsy in young people: with age, the kcc/KCC2 eventually achieves a normal ionic balance and proper inhibitory GABA signalling [54].
The Nhe3 model may facilitate further research on the development of ASD in young brains as well as the development of early biomarkers and treatments. Consistency between the fly and human datasets at both ages indicates a modification of a fundamental sensory mechanism comprising two components that have been conserved over 500 million years of evolution. The conservation of the phenotype and mechanisms from fly to human opens up the option to use the unrivalled genetic tractability of the fly to dissect the molecular mechanisms underpinning the disorder.

Methods (a) Drosophila stocks
Two Drosophila melanogaster genotypes were used as ASD models. The Nhe3 loss-of-function P-element insertion (Nhe3 KG08307 homozygotes) mutation was homozygous PfSUPor-PgNhe3 KG08307 (Bloomington Drosophila Stock Center (BDSC) 14715). The deficiency was Df(2L)BSC187 (BDSC 9672). To avoid second site mutations in the P-element stock, we used the hemizygote Nhe3 KG08307 /Df(2L)BSC187 as a second experimental genotype. For our control cross, we mated the laboratory stock of Canton-S (CS) flies with those with isogenic chromosomes 2C and 3 J [55]. All tested flies had dark red eyes. All genotypes were raised in glass bottles on yeast-cornmeal-agar-sucrose medium (10 g agar, 39 g cornmeal, 37 g yeast, 93.75 g sucrose per litre). They were kept at 258C on a 12 L : 12 D cycle. Male flies were collected on CO 2 the day after eclosion and placed on Carpenter [56] medium in the same environmental conditions for either 3 days or 14 days. Flies were tested approximately between the 4th and 9th hour of the daylight cycle.
(b) Drosophila electroretinography ssVEPs were obtained from the fruit flies [20,23]. Flies were recorded in pairs in a dark room. They were placed in small pipette tips and secured in place with nail varnish. One glass saline-filled electrode was placed inside the proboscis of the fly and another on the surface of the eye. A blue (467 nm wavelength) LED light (Prizmatix FC5-LED) with a Gaussian spectral profile (FWHM 34 nm) was placed in front of the flies together with a diffuser screen and used for temporal contrast stimulation. Flies were dark adapted for at least 2 min and then tested for signal quality with six light flashes. Steady-state stimulation lasted 12 min and comprised seven contrast levels (0-69% in linear steps), each with five repetitions. The frequency of the light flicker was 12 Hz. Each trial (contrast level repetition) was 11 s. The order of the contrast conditions was randomized. The stimulation and the recording from the fly were controlled by in-house MATLAB scripts (scripts can be found in https:// github.com/wadelab/flyCode).

(c) Adult electroencephalography
One hundred neurotypical adult participants (32 males, mean age 21.87, range 18 -49, no reported diagnosis of ASD, reportedly normal or corrected-to-normal vision) took part in the autism spectrum quotient (AQ) measurement study. The AQ is an instrument used for quantifying autistic traits in the neurotypical population and has been shown to have high face validity and reliability in these populations [33]. Owing to time constraints, we used an abridged version of the AQ questionnaire which consists of 28 questions rather than the typical 50 (AQ-Short) [57]. Scores were then scaled to fit the conventional AQ scale. Each participant completed the AQ questionnaire on a computer in the laboratory. The participants were then median split (median ¼ 14) into high and low AQ groups.
For the autistic adult ssVEP study, 12 typical-IQ autistic participants and 12 gender-and age-matched controls (11 males, mean age 23.53, range 18 -39, reportedly normal or corrected to normal vision) took part. ASD diagnosis was confirmed with the Autism Diagnostic Observation Schedule, second edition (ADOS-2). Although IQ was not explicitly measured in this study, all adults had normal speech and a high level of independence (the majority were university students). The absence of ASD diagnosis in the neurotypical participants was also confirmed with ADOS-2 (none of the control participants met criteria for ASD).
Steady-state VEPs were recorded using an ANT Neuro system with a 64-channel Waveguard cap. EEG data were acquired at 1 kHz and were recorded using ASALab, with stimuli presented using MATLAB. The timing of the recording and the stimulation was synchronized using 8-bit low-latency digital triggers. All sessions were performed in a darkened room, testing lasted 45-60 min with approximately 20 min set up time.
Stimuli were presented on a ViewPixx display (VPixx Technologies Inc., Quebec, Canada) with a mean luminance of 51 cd m 22 and a refresh rate of 120 Hz. Stimuli were 0.5 cycle deg 21 sine-wave gratings enveloped by a raised cosine envelope. Gratings subtended 3 degrees of visual angle and were tiled in a 17 Â 9 grid. The participants fixated on a circle in the middle of the screen and performed a fixation task (two-interval-forcedchoice contrast discrimination) to maintain attention. All participants were able to perform the task at above chance levels. There were seven contrast conditions for the flickering gratings (0%, and 2-64% in logarithmic steps, where C % ¼ 100(L max 2 L min )/(L max þ L min ), L is luminance) and eight repetitions. Stimuli flickered on/ off sinusoidally at 7 Hz. Trials were presented in random order in four testing blocks with short breaks in between. Each trial was 11 s long and contained gratings of a random spatial orientation to avoid orientation adaptation effects. These trials were intermixed with orthogonal masking trials that are not presented as part of this study. Data were taken from the occipital electrode Oz.

(d) Child electroencephalography
Thirteen children with a diagnosis of ASD and 20 neurotypical controls matched on gender ratio (10 and 12 males, respectively) and average age (mean age 9.31 and 8.94, respectively, range 5 -13) completed the study. Three of the neurotypical children were tested but excluded due to having autistic siblings (17 participants were included). All children were in mainstream local schools (if they were old enough) and did not have other (or any, in the case of the neurotypical group) reported history of serious medical, psychiatric or neurological conditions. Steady-state EEG data were acquired with a 128-channel HydroCell Geodesic Sensor Net (Electrical Geodesics Inc.). Data were digitized at 432 Hz and band-pass filtered from 0.3 Hz to 50 Hz and were recorded using NETSTATION 4.3 software. Highly noisy data were excluded by removing repetitions with amplitudes that were four standard deviations away from the group mean (for each contrast level and harmonic individually). There were 10 repetitions in total; however, two autistic and one neurotypical child only completed eight repetitions.
Increasing contrast sweep ssVEPs were used. Stimuli for this experiment were presented on an HP1320 CRT monitor with 800 Â 600 pixel resolution, 72 Hz refresh rate and mean luminance of 50 cd m 22 . Stimuli were random binary noise patterns of two luminance levels that increased by contrast in nine logarithmic steps (0-50%) of 1 s each. Each trial contained a prelude at the initial value of the sweep and a postlude at the final sweep value, lasting 12 s in total. Stimuli flickered at 5.12 Hz. Data from the middle 9 s during the sweep were binned according to contrast steps. Methodological differences between the adult and child datasets were due to different conventions being used by the two laboratories in which data were collected.

(e) Data analysis
A fast Fourier transform (in MATLAB) was used to retrieve steadystate response amplitudes at the stimulation frequency (12 Hz for fruit flies, 7 Hz for adult participants and 5.12 Hz for children) and at the second harmonic (24 Hz, 14 Hz and 10.24 Hz, respectively). Fourier transforms were applied to 10 s of each trial (first 1 s discarded; total trial length was 11 s) for the fruit fly and the adult participant datasets and to 1 s binned data for the children's dataset. Contrast response functions were obtained by coherently averaging the amplitudes over repetitions for each contrast level within a participant. Group/genotype scalar means over response amplitude (discarding phase angle) were then calculated for each contrast across participants/flies. Two-way (harmonicÂgroup) ANOVAs were performed on amplitudes at the highest contrast level to investigate the interactions and group effects in all human datasets where only two groups were compared. To identify at which harmonic the autistic children showed a decreased response, two independent samples t-tests were also conducted. One-way ANOVAs with simple planned contrasts were conducted to assess the genotype differences in fruit fly first and second harmonic responses separately as that aided the interpretability of the results between the three genotypes.
To investigate the dissociation between first and second harmonic responses, a scaled ratio (1F 2 2F)/(1F þ 2F) (where 1F is the first and 2F is the second harmonic) was calculated for each participant/fly and each contrast condition. To increase the power of statistical analyses and to decrease the type I error rate, the ratios were then averaged over the contrast conditions that had first harmonic amplitudes significantly above the baseline response (0% contrast condition). For fruit flies, this was six conditions (11.5 -69%); for adult participants, this was four conditions (8-64%); and for children, this was five conditions (8.5 -50%). This procedure resulted in a single frequency-ratio index for each participant/fly. One-way ANOVAs with simple planned contrasts (comparing mutant genotypes with the control genotype) were conducted on the fly frequency ratios for each age separately. Independent t-tests were used to compare frequency ratios in all human datasets between groups. Additionally, a linear regression was conducted on the adult AQ measurement dataset to assess the predictive power of AQ scores on the ratios between frequencies. All statistical tests were two-tailed.
Ethics. All participants in the study gave informed consent and were debriefed on the purpose of the study after the experiment. The experiments were approved by the Department of Psychology Ethics Committee at the University of York.
Data accessibility. Data are available as part of the electronic suplementary material.