As a part of this study, 352 older individuals were screened for potential participants. The inclusion criteria were a score ≥ 4 in Mini-Cog and less than or equal to 25 on the Quick Neurological Screening Test (QNST) to recruit cognitively and neurologically normal individuals for the study. Individuals with a history or self-reported psychological disorders, uncorrected visual impairment, and any other causes of hearing loss other than age were excluded from the study. These individuals were assessed for their hearing thresholds using standard pure-tone audiometry and SIN-K (speech in noise test - Kannada) (Avinash et al. 2010).

Individuals with mild to moderate hearing loss were grouped into the hearing loss group (n = 20), and age, and education-matched individuals with less than minimal hearing loss were grouped into the normal hearing group (n = 20). All the participants gave written consent before the test procedure. Table 1 below shows the demographic and participant characteristics of the participants in both groups. Figure 1 represents the mean pure tone thresholds of both the groups with 1SD.

Mean pure tone thresholds for the right (red) and left (blue) ears of hearing loss and normal hearing group

The detailed assessment encompassed cognitive tests with simultaneous EEG Recording. The testing was done in a dimly lit, sound-treated room with comfortable participant seating.

This study used well known Attention Network Test (ANT) in Eprime 3, and the paradigm used is the same as initially described by Fan et al. (2002). The experiment comprised 3 blocks with 96 trials each, totaling 288 trials. The trials were presented randomly (4 cue conditions x 2 stimulus positions x 2 stimulus directions x 3 congruency conditions x 2 repetitions). Each trial had five components: Initial fixation with random variable duration (ranging from 400 to 1600ms), then cue appeared for 100ms, followed by a middle fixation time of 400 ms. Then, the stimulus consisting of target and flankers was presented until the participant responded or for a maximum of 1700ms. Once the participant responds, the stimulus disappears, and there will be a post-stimulus fixation period of variable duration (3500 msec−first fixation duration—participant’s reaction time). Please see Fan et al. (2002) for the detailed paradigm.

The experiment involved displaying a fixation cross at the center of the screen, with cue stimuli appearing above or below the fixation cross (spatial cue), in the center (central cue), or not at all (no cue). The target stimulus consisted of five horizontal arrows or lines above or below the fixation cross. Each arrow or line was 0.55° of visual angle in length, and the distance between adjacent arrows or lines was 0.06° visual angle. The stimuli, which included a central arrow and four flankers, totaled 3.08° visual angle. These stimuli were presented either 1.06° above or below the fixation point to introduce the attentional-orienting component. The target location was always uncertain except when a spatial cue was presented.

During the test, participants were seated 65 cm from the computer screen to maintain a consistent visual angle of the stimulus. They were instructed to focus on the fixation point at the center throughout the test. All participants were asked to quickly respond to the direction of the center arrow (target) while ignoring the surrounding arrows (flankers) using the response device (Chronos). Initially, the participants completed a practice block of 24 trials, and accuracy and reaction time feedback was provided before the experimental blocks. Figure 2 demonstrates the block diagram of the ANT paradigm, and Table 2 describes the cue condition.

The Attention Network Test is a tool designed to assess the efficiency of three distinct attention networks in the brain: alerting, orienting, and executive control. Each network plays a critical role in processing information and guiding behavior. The alerting network is responsible for achieving and maintaining a state of heightened alertness. It was measured by presenting cues that signal the appearance of a target stimulus. The difference in reaction time between trials with and without a cue assesses the efficiency of the alerting network. A faster response to cued trials indicates a more efficient alerting network.

The orienting network selects sensory input information, directing attention to specific locations or modalities. This was tested by presenting cues that indicate the location where the target will appear (above or below the fixation). The difference in reaction times between correctly cued and centrally cued or uncued trials measures the orienting network’s efficiency. An efficient orienting network is indicated by quicker responses to correctly cued trials.

The executive control network manages complex cognitive processes, including resolving conflict among responses. This was assessed through trials that presented conflicting information, such as incongruent flankers surrounding the target. The difference in reaction time between congruent (non-conflicting) and incongruent (conflicting) trials was used to gauge the efficiency of the executive control network. A smaller difference suggests a more efficient executive control network capable of swiftly resolving conflicts.

Note: The duration of each component in the trial is mentioned in milliseconds in the above figure

Block diagram of the ANT paradigm

A task-related EEG was recorded using an active channel amplifier with 32 electrodes placed based on a 10–20 system (ActiChamp, Brain Vision v005 10/2017). A continuous EEG with Cz as a reference was recorded with an online filter from 1 to 100 Hz and a sampling frequency of 500 Hz. Every participant actively performed the ANT experiment during the EEG Recording. Electrode impedance was maintained below 50 kΩ throughout the testing.

The EEG data were preprocessed using EEGLAB v2021.0 (Delorme and Makeig 2004) running on Matlab R2018b. EEGLAB is a freely available academic software used to process electrophysiological data. EEG lab helps to process high-density EEG and other brain data through its interactive graphic user interface. The standard preprocessing pipeline (shown in Fig. 3) was followed for this study. Based on visual inspection, bad blocks were marked for exclusion from further analysis. Channels were re-referencing to average reference, and the recording reference channel (Cz) was included in the analysis. EEG signals were bandpass filtered between 1 and 30 Hz and then epoched between − 800 and 1000 msec in ERPLAB version 8.10 (Lopez-Calderon and Luck 2014).

Standard preprocessing pipeline for EEG data analysis

Further, the EEG data were decomposed using Independent Component Analysis (ICA) to separate the nonbrain activities. After which, each decomposed component was classified using an automatic independent component classifier called “ICLables.” Based on the labeling done by ICLables, components like blinks, line noise, muscle artifacts, EKG, and other nonbrain activities were marked for rejection and then removed manually. Following this, ERP was computed by averaging across the trials and loaded to a measurement tool in ERPLAB to quantify N100 and P300 components. The P300 component is a positive peak following N2 in the latency range of 300–600ms. N100 is the first negative component after stimulus onset in the 80−120 ms latency range. The mean amplitude is the average amplitude in microvolt in the defined latency range. Peak Latency is the latency of amplitude maxima in the defined latency range.

Behavioral responses were evaluated based on mean reaction time and accuracy and derived network scores representing differences between conditions. Specifically, the alerting network score was calculated as the reaction time difference between double cue and no cue conditions, the orienting network score as the reaction time difference between spatial cue and center cue conditions, and the executive control network score as the reaction time difference between incongruent and congruent conditions. As per the literature, the main aim of ANT is to infer the triple networks of attention as per the Posner model of the attention network. Hence, the current study mainly highlights the network function, while overall statistics is given for the raw scores.

Further, the electrophysiological equivalent of attention networks (Ma et al. 2023; Neuhaus et al. 2010) is calculated as alerting network = target locked N100 in no cue versus double cue condition; orienting network = target locked N100 in the center versus spatial cue condition finally, executive network = overall P300 amplitude in congruent versus incongruent condition. In the current study, the parietal electrodes P3 and P4 were used to measure alerting and orienting attention, while the electrode Pz was used to measure executive control attention. These electrodes are known to show significant changes in early attention and executive function, as reported by Luck (2012), Mangun & Hillyard (1988), Ma et al. (2023), and Neuhaus et al. (2010). Following the stimulus presentation, the study analyzed the average amplitude of N100 and P300 at 150–250ms and 300–600ms, respectively (Ma et al. 2023).

All data was analysed using IBM SPSS Statistics 22. The normality of the data was checked using the Shapiro-Wilk test. As the data was found to be normally distributed, the following statistical tests were carried out: An independent t-test was used to compare the groups in each attentional network score. For electrophysiological measures, a 2 × 2 × 2 repeated measures ANOVA was used to analyze alerting and orienting networks, and a 2 × 2 repeated measures ANOVA was used to analyze the executive control network. All post-hoc analyses were conducted, applying Bonferroni corrections to account for multiple comparisons. A Pearson product correlation coefficient was used to correlate PTA, SNR loss, and attentional network scores.