Based on the Revised Speech Perception in Noise Test (Bilger et al, 1984; Kalikow et al, 1977), the British English Semantic Sentence Test (BESST) stimuli are produced by a British English native speaker. Lists of 22 sentences (11 high-predictability and 11 low-predictability) were presented in the freefield in a background of speech-modulated noise with the same long-term average spectrum as the target speech (Knight, 2017), at a fixed signal-to-noise ratio (SNR) of -1 decibel (dB) SNR. Two practice sentences (one high- and one low-predictability) were presented to participants at a slightly more favourable fixed SNR (2 dB SNR) prior to commencing the main test. Participants were asked to listen to each sentence and repeat the last word aloud. The task was scored as the percentage of last words correctly repeated for both high-predictability and low-predictability lists. Higher scores indicate better performance.

The MCRM is a measure of target talker speech perception in the presence of another talker, presented at an adaptive signal-to-noise ratio (SNR). Participants were presented with sentences in the form of 'show the [animal] where the [color] [number] is'. Participants were asked to listen for the color and number spoken by the female speaker ('dog' was always the animal target) while ignoring the male speaker, and respond by pressing the corresponding target color number on a touchscreen computer. The test used an adaptive 1-up 1-down staircase method and continued until a total of eight reversals were achieved. The test was completed twice by each participant and a third time for instances where there was a difference of ≥5 decibel (dB) SNR between participants' first two test scores. Speech reception thresholds were calculated using the average of the last two reversals, averaged across the 2 or 3 tests. Lower scores indicate better performance.

A subtest from the Wechsler Adult Intelligence Scale, Third Edition (WAIS-III) (Wechsler, 1997), this task involved listening to a string of numbers of increasing length and repeating them in reverse order. The test was presented using pre-recorded digits delivered using the Medical Research Council (MRC) Institute of Hearing Research System for Testing Auditory Responses (IHR-STAR) platform via a speaker situated directly in front of the participant. Trials began with strings of two numbers, finishing at strings of eight. Each string length was presented twice. Strings increased in length by one digit if participants correctly recalled one of the two digit strings at each length, otherwise the test was discontinued. Task performance was scored as the total number of strings correctly recalled in reverse order (of a maximum 14 trials). Higher scores indicate better performance.

The Test of Attention in Listening (TAIL) is a measure of auditory attention using tones that vary in both frequency and spatial location (Zhang et al., 2012). Primary tasks included both frequency and location discrimination and participants were asked to respond as to whether two tones were the 'same' or 'different' in terms of either frequency or location . Tones were presented in the freefield at participants' most comfortable loudness (MCL) level (Ventry et al., 1971) via two speakers situated at 90° to the left and 90° to the right of the participant at a distance of 50cm. TAIL measures the ability to focus selectively on a task-relevant dimension (either frequency or location) and ignore information from task-irrelevant dimensions, using reaction time (RT) as the primary performance measure. This test produced two measures per condition (frequency and location), calculated from the RT data; Distraction and Conflict Resolution. Lower scores indicate better performance.

The Test of Everyday Attention (TEA) subtests 6 (telephone search) and 7 (telephone search while counting) assess single (visual) attention and dual (auditory and visual) attention (Robertson, Ward, Ridgeway, & Nimmo-Smith, 1994). For subtest 6, participants were asked to search a telephone directory for matching symbols. For subtest 7, participants were asked to search a telephone directory for matching symbols while counting strings of beeps in varying lengths (2 to 12) presented in the freefield. The task as scored using time (seconds) per correctly identified symbol, weighted in subtest 7 by the proportion of correctly counted beep strings. A dual task decrement was calculated as the difference in time (in seconds) per correctly identified symbol where two simultaneous tasks are being completed, compared with that for a single task (subtest 7 minus subtest 6). Lower scores indicate better performance.

The size-comparison span (SICSPAN) task (Sorqvist, Ljungberg, & Ljung, 2010) asked participants to view lists of size comparisons (e.g. 'tree is larger than acorn') and respond 'yes' or 'no' using a button box. Participants were then provided with to-be-remembered words from the same semantic category (e.g. 'leaf'). At the end of the list, participants were required to recall the to-be-remembered words while inhibiting words included in the size comparison judgments. The task began with lists of 2 size comparison judgments and to-be-remembered words, increasing to list lengths of 3, 4, 5 and 6. There were two trials at each list length. The task continued until all list lengths had been presented, with no discontinuation rule. The task was scored using the number of list items correctly recalled (maximum 40). A second measure (intrusions) recorded the number of incorrect items recalled, or errors in inhibition, for each participant. Higher scores indicate better performance.

The dual task is a measure of listening and memory designed to assess listening effort (Howard, Munro, & Plack, 2010). Participants were presented with a 5-digit memory task that flanked a speech-in-noise comprehension task. A string of 5 digits was displayed visually on a computer screen for 5 seconds. Participants were asked to retain the digits in memory for later recall. Participants were then presented with a list of 5 Arthur Boothroyd (AB) isophonemic monosyllabic words (Boothroyd, 1968) presented in multi-talker babble, and were asked to repeat each word immediately after presentation. After each list of five words, participants were asked to recall the previously presented five digits. There were 4 word lists, resulting in a maximum possible score of 20 correctly repeated words and 20 correctly recalled digits. A dual-task score was calculated by adding together the scores for the word and digit tasks (maximum 40). Higher scores indicate better performance.

The Hearing Handicap Inventory for the Elderly is a 25-item validated questionnaire that quantifies the emotional and social/situational effects of self-perceived hearing impairment to quantify hearing-specific quality of life. Participants were asked to complete the 25-item paper questionnaire answering statements such as 'Does a hearing problem cause you to be nervous' with either 'yes' (4 points), 'no' (0 points) or 'sometimes' (2 points). The questionnaire was scored as total points for all items (maximum 100 points). Subtotal scores can also be derived for emotional (12 items, maximum 48 points) and situational item subscales (13 items, maximum 52 points). Lower scores indicate better performance.

The Glasgow Hearing Aid Benefit Profile (GHABP) is a validated questionnaire to assess self-reported activity limitations (GHABP Initial Disability) and participation restrictions (GHABP Handicap) arising from difficulties in hearing, as well as hearing aid use, benefit, and satisfaction (Gatehouse, 1999). For the purposes of the present study population (experienced hearing aid users) only the first two measures were assessed. Participants were presented with four listening scenarios and were asked to rate the amount of difficulty they have in that situation while wearing their hearing aids (1 = no difficulty to 5 = cannot manage at all), together with how much any difficulty either worried, annoyed or upset them ( 1 = not at all to 5 = very much indeed). The mean of all four scenarios in each measure were converted to a percentage score for GHABP Initial Disability and GHAPB Handicap. Lower scores indicate better performance.

Online training intervention: A non-adaptive placebo version of Cogmed RM working memory training. Tasks are fixed at a low-difficulty practice level (three to-be-remembered items) and do not increase in difficulty over the course of the intervention. Participants complete 35-45 minutes of active placebo training per day, 5 days a week for 5 weeks.

Cogmed RM - Online (non-adaptive, placebo) working memory training intervention

Total of all reporting groups

Race and Ethnicity were not collected from any participant.

Scores on the MoCA range from zero to 30, with a score of 26 and higher generally considered normal cognitive functioning.

Better ear pure tone audiometric thresholds measured in decibels (dB) averaged across octave frequencies 0.25, 0.5, 1k, 2k and 4k Hz.

On the decibel scale, the quietest audible sound (perceived near total silence) is 0 dB. A sound 10 times more powerful is 10 dB. Lower pure-tone audiometric thresholds indicate better hearing.

The phoneme discrimination task assesses an individual's ability to distinguish differences between phonemes presented on a continuum. Participants were presented with three discrete phonemes per trial from a continuum of 96 sound files (48 for each phoneme within a pair). For each trial, two of the phonemes were identical and one was different. Participants were asked to identify the odd one out. Two different phoneme pairs: /a/ /e/ (easy) and /d/ /g/ (difficult) were presented for a block of 35 trials in sequential blocks, with a 3-trial demonstration of continuum /a/ /e/. The task was presented in quiet room and difficulty was adapted based on individual participant performance using a three-phase adaptive staircase procedure (Moore, 2005) to provide a phoneme discrimination threshold, which is calculated as the average distance between the 96 sound files for the last 2 reversals in a block of 35 trials. Scores range from 50-100, with lower scores indicating better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

Based on the Revised Speech Perception in Noise Test (Bilger et al, 1984; Kalikow et al, 1977), the British English Semantic Sentence Test (BESST) stimuli are produced by a British English native speaker. Lists of 22 sentences (11 high-predictability and 11 low-predictability) were presented in the freefield in a background of speech-modulated noise with the same long-term average spectrum as the target speech (Knight, 2017), at a fixed signal-to-noise ratio (SNR) of -1 decibel (dB) SNR. Two practice sentences (one high- and one low-predictability) were presented to participants at a slightly more favourable fixed SNR (2 dB SNR) prior to commencing the main test. Participants were asked to listen to each sentence and repeat the last word aloud. The task was scored as the percentage of last words correctly repeated for both high-predictability and low-predictability lists. Higher scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The MCRM is a measure of target talker speech perception in the presence of another talker, presented at an adaptive signal-to-noise ratio (SNR). Participants were presented with sentences in the form of 'show the [animal] where the [color] [number] is'. Participants were asked to listen for the color and number spoken by the female speaker ('dog' was always the animal target) while ignoring the male speaker, and respond by pressing the corresponding target color number on a touchscreen computer. The test used an adaptive 1-up 1-down staircase method and continued until a total of eight reversals were achieved. The test was completed twice by each participant and a third time for instances where there was a difference of ≥5 decibel (dB) SNR between participants' first two test scores. Speech reception thresholds were calculated using the average of the last two reversals, averaged across the 2 or 3 tests. Lower scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

A subtest from the Wechsler Adult Intelligence Scale, Third Edition (WAIS-III) (Wechsler, 1997), this task involved listening to a string of numbers of increasing length and repeating them in reverse order. The test was presented using pre-recorded digits delivered using the Medical Research Council (MRC) Institute of Hearing Research System for Testing Auditory Responses (IHR-STAR) platform via a speaker situated directly in front of the participant. Trials began with strings of two numbers, finishing at strings of eight. Each string length was presented twice. Strings increased in length by one digit if participants correctly recalled one of the two digit strings at each length, otherwise the test was discontinued. Task performance was scored as the total number of strings correctly recalled in reverse order (of a maximum 14 trials). Higher scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The Test of Attention in Listening (TAIL) is a measure of auditory attention using tones that vary in both frequency and spatial location (Zhang et al., 2012). Primary tasks included both frequency and location discrimination and participants were asked to respond as to whether two tones were the 'same' or 'different' in terms of either frequency or location . Tones were presented in the freefield at participants' most comfortable loudness (MCL) level (Ventry et al., 1971) via two speakers situated at 90° to the left and 90° to the right of the participant at a distance of 50cm. TAIL measures the ability to focus selectively on a task-relevant dimension (either frequency or location) and ignore information from task-irrelevant dimensions, using reaction time (RT) as the primary performance measure. This test produced two measures per condition (frequency and location), calculated from the RT data; Distraction and Conflict Resolution. Lower scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The Test of Everyday Attention (TEA) subtests 6 (telephone search) and 7 (telephone search while counting) assess single (visual) attention and dual (auditory and visual) attention (Robertson, Ward, Ridgeway, & Nimmo-Smith, 1994). For subtest 6, participants were asked to search a telephone directory for matching symbols. For subtest 7, participants were asked to search a telephone directory for matching symbols while counting strings of beeps in varying lengths (2 to 12) presented in the freefield. The task as scored using time (seconds) per correctly identified symbol, weighted in subtest 7 by the proportion of correctly counted beep strings. A dual task decrement was calculated as the difference in time (in seconds) per correctly identified symbol where two simultaneous tasks are being completed, compared with that for a single task (subtest 7 minus subtest 6). Lower scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The size-comparison span (SICSPAN) task (Sorqvist, Ljungberg, & Ljung, 2010) asked participants to view lists of size comparisons (e.g. 'tree is larger than acorn') and respond 'yes' or 'no' using a button box. Participants were then provided with to-be-remembered words from the same semantic category (e.g. 'leaf'). At the end of the list, participants were required to recall the to-be-remembered words while inhibiting words included in the size comparison judgments. The task began with lists of 2 size comparison judgments and to-be-remembered words, increasing to list lengths of 3, 4, 5 and 6. There were two trials at each list length. The task continued until all list lengths had been presented, with no discontinuation rule. The task was scored using the number of list items correctly recalled (maximum 40). A second measure (intrusions) recorded the number of incorrect items recalled, or errors in inhibition, for each participant. Higher scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The dual task is a measure of listening and memory designed to assess listening effort (Howard, Munro, & Plack, 2010). Participants were presented with a 5-digit memory task that flanked a speech-in-noise comprehension task. A string of 5 digits was displayed visually on a computer screen for 5 seconds. Participants were asked to retain the digits in memory for later recall. Participants were then presented with a list of 5 Arthur Boothroyd (AB) isophonemic monosyllabic words (Boothroyd, 1968) presented in multi-talker babble, and were asked to repeat each word immediately after presentation. After each list of five words, participants were asked to recall the previously presented five digits. There were 4 word lists, resulting in a maximum possible score of 20 correctly repeated words and 20 correctly recalled digits. A dual-task score was calculated by adding together the scores for the word and digit tasks (maximum 40). Higher scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The Hearing Handicap Inventory for the Elderly is a 25-item validated questionnaire that quantifies the emotional and social/situational effects of self-perceived hearing impairment to quantify hearing-specific quality of life. Participants were asked to complete the 25-item paper questionnaire answering statements such as 'Does a hearing problem cause you to be nervous' with either 'yes' (4 points), 'no' (0 points) or 'sometimes' (2 points). The questionnaire was scored as total points for all items (maximum 100 points). Subtotal scores can also be derived for emotional (12 items, maximum 48 points) and situational item subscales (13 items, maximum 52 points). Lower scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

The Glasgow Hearing Aid Benefit Profile (GHABP) is a validated questionnaire to assess self-reported activity limitations (GHABP Initial Disability) and participation restrictions (GHABP Handicap) arising from difficulties in hearing, as well as hearing aid use, benefit, and satisfaction (Gatehouse, 1999). For the purposes of the present study population (experienced hearing aid users) only the first two measures were assessed. Participants were presented with four listening scenarios and were asked to rate the amount of difficulty they have in that situation while wearing their hearing aids (1 = no difficulty to 5 = cannot manage at all), together with how much any difficulty either worried, annoyed or upset them ( 1 = not at all to 5 = very much indeed). The mean of all four scenarios in each measure were converted to a percentage score for GHABP Initial Disability and GHAPB Handicap. Lower scores indicate better performance.

Outcome measured at 0, 2, 7 and (adaptive training group only) 31 weeks. Training-related improvements in outcomes are assessed using the change in performance between pre-training (2 weeks) to post-training (7 weeks) measures.

Online training intervention: A non-adaptive placebo version of Cogmed RM working memory training. Tasks are fixed at a low-difficulty practice level (three to-be-remembered items) and do not increase in difficulty over the course of the intervention. Participants complete 35-45 minutes of active placebo training per day, 5 days a week for 5 weeks.

Cogmed RM - Online (non-adaptive, placebo) working memory training intervention