Gamma-Range Auditory Steady-State Responses and Cognitive Performance: A Systematic Review

The auditory steady-state response (ASSR) is a result of entrainment of the brain’s oscillatory activity to the frequency and phase of temporally modulated stimuli. Gamma-range ASSRs are utilized to observe the dysfunctions of brain-synchronization abilities in neuropsychiatric and developmental disorders with cognitive symptoms. However, the link between gamma-range ASSRs and cognitive functioning is not clear. We systematically reviewed existing findings on the associations between gamma-range ASSRs and cognitive functions in patients with neuropsychiatric or developmental disorders and healthy subjects. The literature search yielded 1597 articles. After excluding duplicates and assessing eligibility, 22 articles were included. In healthy participants, the gamma-range ASSR was related to cognitive flexibility and reasoning as measured by complex tasks and behavioral indicators of processing speed. In patients with schizophrenia, the studies that reported correlations found a higher ASSR to be accompanied by better performance on short-term memory tasks, long-term/semantic memory, and simple speeded tasks. The main findings indicate that individual differences in the gamma-range ASSR reflect the level of attentional control and the ability to temporary store and manipulate the information, which are necessary for a wide range of complex cognitive activities, including language, in both healthy and impaired populations.


Introduction
Understanding the role of temporal brain synchronization in cognitive functioning has been a subject of intense research for many years [1,2]. Brain electrophysiological measures, such as electroencephalography (EEG) and magnetoencephalography (MEG), provide cost-effective, non-invasive techniques to explore the occurrence of neural synchronization and temporal states [3,4]. In this perspective, attention has been drawn to gamma-range  activity due to its association with cognitive performance [5,6]. A large range of cognitive functions, such as information-processing speed, working memory, abstract reasoning, and verbal abilities [7], which are related to gamma activity [8,9], are impaired in patients with neuropsychiatric and developmental disorders like schizophrenia [10], Alzheimer's disease [11,12], dyslexia [13], aging [14], etc. Along with the observed dysfunctions in cognitive performance, the impaired gamma oscillations were also frequently reported in different patient groups [14,15].
One of the EEG/MEG techniques used to explore individual differences in neural synchronization is the method of auditory steady-state response (ASSR) [16]. The ASSR is an electrophysiological response of the brain that synchronizes to the frequency and phase of rapid, periodic auditory stimuli delivered in trains of clicks [17], broad-band noise [18], amplitude-modulated (AM) tones [19], or sounds modulated with chirps [20,21]. The response to auditory stimulation reaches the greatest magnitude with the presentation of included papers was also conducted to identify potentially relevant reports. All titles and abstracts were scanned for selection criteria. When the abstract provided insufficient information, the methods section of the article was reviewed. The selection-procedure flowchart is presented in Figure 1. The included studies were checked by the first and the last authors. When a disagreement arose, the second author's opinion was sought.

Study Selection
The following inclusion criteria for study selection were used: (1) the participants were adults ≥ 18 years old; (2) EEG/MEG methods with gamma-range  Hz) auditory stimulation were used; (3) a behavioral assessment of cognitive performance was performed (4); a statistical association between ASSR measures and cognitive performance was reported; and (5) the article reported original research. Since this is the first systematic review of the cognitive correlates of gamma-range ASSRs, to be as inclusive as possible, studies in various neuropsychiatric and developmental disorders (e.g., schizophrenia, Alzheimer's disease, dyslexia, etc.) were included. Original articles that were not found in the specified search but were cited or recommended by selected studies, or were known by the review authors and met the inclusion criteria, were included as well. The following papers were excluded: (1) animal studies; (2) studies measuring ASSRs in frequencies other than gamma-range; (3) studies not using adequate cognitive evaluation methods, i.e., not using cognitive tests/tasks or neuropsychological assessment tools; (4) studies in which ASSRs were collected during altered states (e.g., during high-cognitive-demand tasks, sleep, anesthesia, or hallucinations); (5) studies in which ASSRs could be affected by brain-stimulation techniques (e.g., tACS, TMS); (6) papers published in non-English languages. When papers were not accessible as a full-text version or lacked necessary information, efforts were made to retrieve the missing data by contacting the authors.

Data Extraction
For each study, the following information was extracted (Table 1): (1) sample (type, size, age, and gender composition); (2) neurocognitive assessment method (i.e., tasks that were used to assess cognitive performance); (3) auditory stimulation settings (stimulation frequencies, type, number of repetitions, duration); (4) the EEG/MEG assessment (measure, site, latency); and (5) the correlation between measure(s) of ASSRs and neurophysiological measure(s). To systematize the results, we grouped the neurocognitive performance assessment tasks that were used in the included studies into higher-order cognitive domains ( Table 2): composite measures of global cognitive functioning, attentional control and executive functions, short-term and working memory, cognitive flexibility and reasoning, language abilities, and motor abilities. It is important to note that this classification is by no means an exhaustive list of well-established and validated assessment tools for listed cognitive functions, but only an attempt to systematize the methods used in included studies and overcome the disparity in the assessment tools that were used.     ERSP-event-related spectral perturbation, ITC-inter-trial coherence, ITPC-inter-trial phase coherence, MTP-mean trial power, PLI-phase-locking index, PLF-phase-locking factor, SNR-signal-to-noise ratio; n/a-not available; n.s.-not significant.

Quality Evaluation
The quality of included articles was rated in line with guidelines by the Cochrane handbook for systematic reviews of interventions [77] by first and last authors. When a disagreement arose, the second author's opinion was sought. This assessment is primarily focused on the reporting bias with respect to key aspects of the study from the perspective of its reproducibility and replicability. The scale was adapted to capture major sources of bias, and each study was rated depending on the amount and quality of information that was provided in the article (see Supplementary Material).

Results
The literature search yielded 1597 articles. After excluding duplicates and studies that did not meet the inclusion criteria, 22 articles were included in this systematic review ( Figure 1). Twelve out of 22 studies defined the ASSR association to the cognitive correlates as one of the primary study purposes; the remaining 10 reports presented ASSR-cognitive correlates as a non-primary topic.
The EEG/MEG response to repetitive click stimulation in the gamma frequency range was assessed in the majority of included studies; however, variations in stimulation duration, stimulus features, and inter-stimulus interval settings, as well as acquisition methods, were detected (Table 1). Several studies used amplitude-modulated sounds [60,62,75], or chirp-based stimulation [58,68]. The main ASSR outcome measures were power and phase synchronization/consistency; two studies provided signal-to-noise ratio evalua-tions [60,62], and one study provided individual gamma peak frequencies [58]. Most studies focused on the evaluation of the response during the entire stimulation duration [ [58,68]. Three of the studies utilized MEG recordings [68,70,75]. EEG recordings in the selected reports were either performed with the nose serving as a reference, or electrodes were average referenced [44,63,[65][66][67]72,76]. The EEG results were reported mostly for fronto-central locations, with the exception of van Deursen et al. and Hirtum et al., who analyzed temporal and temporoparietal locations [55,62].
Overall, the majority of the included studies were characterized by a low risk of reporting bias, with the exception of Bartolomeo et al. [59] and Gaskins et al. [60], who did not report paradigm settings in a sufficient manner.

Correlations between ASSR and Cognitive Performance
The last column in Table 1 shows the correlation outcomes between performance on cognitive tasks and ASSR measures as reported in the included studies. The studies yielded the full range of effect sizes, with correlations from 0.13 to 0.76 (median correlation: 0.43). Still, it should be noted that nonsignificant correlations have seldom been reported in sufficient detail. To increase transparency and allow for future meta-analytical studies, it is advisable that the authors fully report on statistics for nonsignificant effects. The power of most studies was only sufficient to detect correlations higher than 0.50, with the exception of Kirihara et al. [67] and Koshiyama et al. [65][66][67], who performed their analyses on large samples, and thus were able to detect even correlations in the range of 0.10-0.20. Therefore, it is difficult to estimate the real strength of the relationship between ASSR and different cognitive domains.
In healthy participants, the gamma-range ASSR was related to cognitive flexibility and reasoning as measured by complex tasks such as Tower of London [71], Similarities [73], and the Mazes Test [76]. Additionally, the ASSR was related to behavioral indicators of processing speed, i.e., performance on the Trial making test [76] and Symbol coding [74].
Five out of 16 studies that assessed patients with psychotic symptomatology (schizophrenia, schizoaffective disorder, schizotypal personality disorder) showed no relationship between gamma-range ASSR and cognitive performance [59,61,64,69,70]. The studies that reported correlations found a higher ASSR to be accompanied by better performance on short-term memory tasks (such as Digit span and Letter-Number sequencing) [43,[65][66][67]72], tasks tapping speeded access to long term/semantic memory (like Verbal fluency) [63] or simple speeded tasks (like Symbol coding) [44]. However, observations of complex reasoning tasks such as the Mazes test [76], Similarities (form WAIS-III battery) [74], and Tower of London [44] were not very consistent.
In other patient groups, gamma-range ASSRs were indicative of impairment in diseaserelevant cognitive domains. Namely, studies that assessed language abilities in dyslexia reported a negative correlation with phonological awareness (i.e., performance on a Spoonerism task) and phonological fluency (as measured by RAN), as well as literacy and nonword repetition [62,68]. Furthermore, better overall functioning assessed with ADAS-cog was related to higher 40-Hz ASSRs in patients with mild Alzheimer's disease [55]. Patients with multiple sclerosis who performed better on different cognitive tasks from BRB-N tended to respond at higher gamma frequencies [58]. Still, it should be noted that a study assessing bipolar patients found no relationship between ASSR and cognitive performance measured by several WAIS-III subtests [73].

Discussion
Impaired cognitive performance is frequently reported in patients with neuropsychiatric disorders, and is accompanied by aberrant gamma activity [78,79]. As a method of exploring individual differences in the ability to generate and sustain gamma-range activity, EEG/MEG-based auditory steady-state responses (ASSRs) are used [16]. However, the functional relationship between gamma-range ASSRs and cognitive functioning, and their link to individual differences in performance/abilities, is highly unresolved. This review aimed to evaluate the current state of knowledge on the associations between gamma-range ASSRs and cognitive functions as measured by various cognitive tests or batteries. The literature search was carried out in order to collect, systematize, and critically evaluate previous studies that assessed ASSRs within the 30-80 Hz range and various cognitive domains in the same study sample.
Twenty-two articles were included and analyzed in this review. An absolute majority of the studies were performed with an aim of ASSR evaluation in clinical populations and used wide array of tasks to tap into different cognitive functions. For comparison purposes within this review, the cognitive tasks used in the included studies were grouped into higher-order cognitive domains. It is important to point out that most of the studies used several tasks, but measured specific, and often narrow, cognitive domains. Apart from Rojas et al. [75], who used an abbreviated WAIS, none of the studies conducted a comprehensive cognitive assessment using either a full WAIS or set of test/tasks that would cover all aspects of cognitive abilities as they are defined by well-established accepted models, e.g., the Cattell-Horn-Carroll theory of cognitive abilities [80][81][82]. However, a comprehensive assessment of cognitive functioning was performed for clinical populations using condition-specific batteries in Sun et al., Murphy et al., Leonhardt et al.,and Bartolomeo et al. [59,69,70,76]. Thus, the quality evidence of the relationship between cognition and ASSRs is currently higher for pathological than for normal functioning. To that point, only a few studies with large enough samples to reliably detect even smaller effect sizes found effects in clinical subjects, but not a healthy control group [64][65][66][67]. Smaller sample sizes tended to artificially inflate correlations, while the fact that most studies related very specific EEG measures to very general cognitive measures theoretically resulted in lower effect sizes. Therefore, based on the current evidence, it is difficult to reliably estimate the strength (or even existence) of the relationship between ASSRs and different cognitive functions.
The most frequent associations of gamma-range ASSRs were identified in the outcomes for the tasks assessing processing speed and short-term/working memory tasks, covering both efficiency and timing aspects of the performance. For example, the Symbol coding task was used in six studies [27,[47][48][49]54,55]; in three of those, performance of the task was positively associated with measures of ASSR in patients with schizophrenia [44,74], multiple sclerosis [58], and healthy controls [74]. Similarly, the Digit span test was employed in seven studies [44,63,68,[72][73][74]76]. However, it was positively associated to gamma-range ASSR measurements in two studies of schizophrenia groups [72,74] and first-degree relatives of patients [72], and negatively in one study of dyslexia [68]. The Letter-Number sequencing task was employed in five reports [43,[64][65][66][67], four of which showed a positive association between task performance in patients with schizophrenia and 40-Hz ASSR measures. These results might imply that individual differences in gamma-range ASSRs reflect the individual differences in the ability to focus attention, and maintain and manipulate the information in short-term memory storage. However, it appears that the associations are evident mostly in patient groups, where short-term/working memory is affected.
Additionally, performance on several tasks tapping cognitive flexibility and reasoning correlated with measures of gamma-range ASSRs [71,74,76] in healthy controls [71,74,76] and patients with schizophrenia [74,76]. However, tasks evaluating cognitive flexibility and reasoning are defined by high versatility, and the functions they assess are intricately covering attentional control/executive functioning and memory processes [83,84]. This makes evaluation of a particular aspect that is contributing to the observed relationship difficult. For example, the moves on the Tower of London task (ToL), as assessed in reviewed papers, were positively correlated to the synchronization levels of gamma-range activity [71]. Cazalis et al. proposed that standard performers on ToL might need to put the higher load on working memory to perform the task, compared to superior performers [85]. Should that be the case, a positive relationship between gamma-range ASSRs and moves in ToL could highlight the working-memory-related aspects of the task. This assumption is indirectly supported by the speeded performance on short-term memory tasks (such as Digit span and Letter-Number sequencing) [43,65,72] and long term/semantic memory tasks (like Verbal fluency) [63], as observed in patients with psychotic symptomology, who display stronger/more precise gamma-range ASSRs.
Furthermore, studies designed to assess the relationship between gamma-range ASSRs and the degree of language impairment in dyslexia reported significant associations with several tests, thus suggesting stronger impairment of ASSRs with debilitated language function [62,68]. These results are consistent with reported links between ASSRs and behavioral outcomes of speech recognition [86,87]. Correlation between ASSR and language impairment may be attributed to the altered temporal sampling [33], but might also be a reflection of the common core functions, such as attentional control and the ability to maintain and manipulate content in short-term memory storage. Namely, language comprehension and production strongly depend on the temporary storage and processing of information, i.e., working memory [88]. This is especially prominent in different language disorders [89]. Therefore, it is plausible that the observed relationship between gamma-range ASSRs and language performance does not reflect differences in linguistic abilities per se, but rather stems from individual differences in a more fundamental ability to process information. To untangle this issue, future studies focusing on language performance should include a behavioral assessment of working or short-term memory to examine if the relationship between gamma-range ASSRs and language performance is function-specific or stands as a reflection of a more general ability. In addition, further research is needed to explore the relationship between language and ASSRs in different populations; the association might be of importance in different neuropsychiatric conditions, since the positive correlation between 40-Hz ASSRs and verbal fluency was observed by Kim et al. in schizophrenia patients as well [63].
Importantly, several studies not included in our review reported the relationship between auditory cognition assessed with a gap-detection task, and the preferred frequency of gamma in response to periodic stimulation [21,90]. Studies evaluated the resonant frequency of gamma activity that was also the focal point of the work included in the review [58]. The estimation of preferred gamma frequency is possible using both classical single-frequency stimulation [91] and as a response to the specific type of periodic stimulation with stimulation frequencies spanning a wide range [21,58]. This approach might be particularly promising in clinical testing [20,92], as it allows fast estimation of the individual properties of networks involved in response that might display associations to cognitive functioning [58]. The individual resonant frequencies within the gamma range were negatively related to the speed on attentional control and executive tasks, as shown in patients with multiple sclerosis [58]. This observation is in agreement with a positive relationship between peak gamma frequencies and working-memory performance over the stages of sedation with anesthesia [93,94]. Altogether, the results advise that the state of neural networks defining individual gamma frequencies may relate to the temporal resolution, and to the individual parameters of information-processing speed and performance efficiency.
Even though it seems that differences in gamma-range ASSRs reflect a central aspect of cognitive functioning, i.e., attentional control and information manipulation in both healthy and clinical populations, the available evidence is far from conclusive. It is important to highlight that the reviewed studies utilized various cognitive tasks in which gamma activity is expected to be involved. However, it was not the direct aim of the studies to explain how auditory cortical responses relate to each of those tasks and to shed the light on cognitive processes involved in those tasks. This may be one of the reasons for the lack of systematic assessment of the cognitive performance in relation to ASSRs. Furthermore, several methodological aspects could have contributed to the discrepancy of obtained results. Though the stimulation parameters for ASSRs were relatively consistent when responses to 40 Hz were assessed, the cognitive testing protocols were substantially diverse in the available body of literature. However, even with the same stimulation settings, and ASSR and cognitive-assessment approaches, different correlational outcomes were reported [43,64,65,73,74]. This suggests that inter-individual subject's characteristics, such as age or gender, might have a moderation-like effect, as these are known to affect both cognitive performance [95,96] and ASSRs [64,97,98]. Importantly, differences in the momentarily state of arousal/attention levels [99,100], prevailing psychopathology [46], medication regimen [101], the general level of global functioning, and the stage of disease in neuropsychiatric patients [56,102] could have affected the relationships. Also, the gammarange ASSR measures could potentially be compromised by myogenic and micro-saccadic activity [103] and may influence some of the results between ASSRs and cognitive processes reviewed in the manuscript. Therefore, future studies should adopt the designs that enable exploring the magnitude of the moderation effects of these variables.

Conclusions
Gamma-range ASSRs were associated with the outcomes of short-term and workingmemory tasks, covering both efficiency and timing aspects of the performance, and with the outcomes on tasks aimed at evaluating processing speed. Additionally, performance on several tasks to tap cognitive flexibility and reasoning correlated with measurements of gammarange ASSRs, indirectly supporting an association with attentional control/executive functioning and memory. Finally, a possible link of ASSRs with results of language-tapping tests was observed in dyslexia. We propose that individual differences in gamma-range ASSRs reflect the level of attentional control and the ability to temporarily store and manipulate the information, which is necessary for a wide range of complex cognitive activities in different clinical populations.
This review points out several important directions for future research. First, the assessment of behavioral effects alongside physiological measures, though indirectly, may provide important insights into the underlying mechanisms behind brain functioning and lead to more informative interpretation of results. Second, the heterogeneity of the methods for cognitive assessment and lack of systematic selection of behavioral tasks make it difficult to directly compare and evaluate effects presented in different studies. Therefore, future research would benefit form more theory-and model-driven selections of the cognitive tasks. Moreover, inclusion of several tasks tapping into different cognitive functions would enable a dissociation between different cognitive functions in relation to ASSRs. Third, exploring the same relationships in different patient groups may uncover the regularity between cognitive and physiological deficits that are common across different disorders. Finally, we would like to encourage studies specifically designed to test the hypothesis that gamma-range ASSRs reflect individual differences in working-memory ability and performance.

Conflicts of Interest:
The authors declare no conflict of interest.