Effects of Severe Bothersome Tinnitus on Cognitive Function Measured with Standardized Tests

Abstract

Neurocognitive tests compared abilities in people with bothersome tinnitus against an age, gender, and education matched normative population. Participants between 18 and 60 years had subjective, unilateral or bilateral, non-pulsatile tinnitus for >6 months, and a Tinnitus Handicap Inventory score of ≥38. Results from a first testing session showed deficits in learning, learning rates, immediate recall of heard words, and use of a serial order encoding strategy. Initial reliance on serial order encoding and later, increased intrusion of incorrect words towards normal levels might indicate a less demanding strategy to compensate for weakness in associative memory for semantic categories.

INTRODUCTION

Tinnitus is a perception of sound without actual acoustic stimulation. Approximately, 40 million people in the United States experience chronic tinnitus and 10 million of them consider it to be a severe problem (Henry, Dennis, & Schechter, 2005). Prior neurocognitive evaluations of patients with bothersome tinnitus indicated deficits in tests of memory and attention, substantiating self-reports of impaired concentration. These cognitive deficiencies possibly arose because phantom noises, like chronic pain, involuntarily garnered attention resources (Møller, 2000) and thereby affected cognitive efficiency (Rossiter, Stevens, & Walker, 2006; Stevens, Walker, Boyer, & Gallagher, 2007). For instance, Hallam and colleagues (Hallam, McKenna, & Shurlock, 2004) found significantly poorer performance on a reaction time task (variable fore-period) that was combined with a verbal fluency task in 43 tinnitus patients compared to 32 controls. These findings indicated that slowed reaction times in tinnitus patients might reflect trouble inhibiting their attention to irrelevant tinnitus signals. Additionally, individuals with tinnitus compared to controls were less proficient in recalling word lists and spatial positions, suggesting interrupted processing during memory tasks (Hallam, et al., 2004). These findings of cognitive interference in tinnitus patients might indicate a failure to stop attending to or habituate to the irrelevant tinnitus signal (Hallam, Rachman, & Hinchcliffe, 1984).

The current evaluation of the neurocognitive consequences of tinnitus employed standardized neuropsychological tests with age and gender matched vetted norms. The first objective was to utilize several verbal tests and contrast findings against clinical norms. Prior studies presented only a single test (Stevens, et al., 2007) or lacked standardization against clinical norms (Hallam, et al., 2004). A second objective was to determine if practice from repeated testing improved performance possibly due to implicitly learned compensatory strategies to ignore cognitive interference from tinnitus.

METHODS

Participants

Fourteen participants with tinnitus provided written informed consent to a study protocol previously approved by the Washington University School of Medicine Human Research Protections¹. Enrolled adult participants were between the ages of 18 and 60 with subjective, unilateral or bilateral, non-pulsatile tinnitus of ≥6 month’s duration, and a score of ≥38 on the Tinnitus Handicap Inventory (THI) (Newman, Jacobson, & Spitzer, 1996). No included patients had a Beck Depression Inventory-II ≥18 (Beck, 1961) or other psychiatric or neurologic disorders. An exclusion criterion for all participants was taking any psychoactive drugs that might have affected cognitive performance. Another exclusion criterion for this study was a history of any drug or alchohol abuse. Table 1 lists demographic and tinnitus characteristics. A description of audiometric deficits is below.

Table 1.

Demographic and tinnitus characteristics.

Demographics
Age	52 (42-59) yrs#
Male	10 (71%)
Female	4 (29%)
White	13 (93%)
Other	1 (7%)
Education	16 (13-18) yrs#
Tinnitus Characteristics
Duration	7.0 (0.5-17.9) yrs#
Loudness	7.5 (5.0-9.0)#
THI score	51 (38-76)#
Interference with sleep
Often	7 (50%)
Sometimes	6 (43%)
No	1 (7%)
Effort to ignore
Constant	4 (29%)
Considerable	6 (43%)
Some	3 (21%)
Slight	1 (7%)
Discomfort
Great	4 (29%)
Very	2 (14%)
Moderate	6 (43%)
Mild	1 (7%)
None	1 (7%)
Sought medical help
Yes	10 (71%)
No	4 (29%)

^#Median and range

Tinnitus Handicap Index, THI

Neurocognitive Tests

Controlled Oral Word Association Test

(COWAT/FAS) (Benton & Hamsher, 1989) measured the ability to generate words beginning with specified letters (i.e., F, A, and S) and, as such, detected verbal association fluency.

California Verbal Learning Test

(CVLT) is a multifactorial verbal test of auditory attention, learning, and memory functions (D. C. Delis, Kramer, Kaplan, & Ober, 1987; Elwood, 1995; Lezak, 2004). Tests of immediate free recall involved hearing two lists (A & B) each composed of 16 different words, pseudo-randomly ordered and equally divided into words from four categories (List A: animals, vegetables, vehicles, and furniture; List B: animals, vegetables, musical instruments, and architectural house features). There were 5 presentations of List A (Trials 1-5) and one presentation for List B immediately after the fifth trial for List A. The number of freely recalled correct words for each trial measured learning and total correct per successive trials across 5 trials for List A indicated learning rate. Memory tests involved free and cued recall of list A tested immediately (short-delay) and again after 20 minutes (long-delay) with the intervening time filled with unrelated tasks. On cued recall trials, participants must recall the list A words for an examiner specified category. A recognition memory task for List A words involved a 44-word list read aloud by the examiner. In this task, the participant had to recognize whether a presented word was one previously learned (e.g., a target) or had to identify the presented word as novel (e.g., a distractor). Some distracters shared semantic categories with target words, while others were phonemically similar.

All participants underwent three neurocognitive assessments with a two week interval between session 1 and a second testing session (session 2) and a 4 week interval before the third testing session (session 3).

Statistical Analyses

For the CVLT, comprehensive scoring software for the standardized tests automatically transformed an individual score (number of correctly recalled words) into a z-score (D.C. Delis & Fridlund, 1987). The z-score transformation was in reference to normative distributions of raw scores that matched the age, gender, and education level of a participant (D. C. Delis, et al., 1987). The metric for transforming total number of words recalled across five repeated learning trials of List A was a standardized T-score in reference to a population mean of 50 and standard deviation of 10 (D. C. Delis, et al., 1987). Wilcoxon Signed Ranks tests (PROC UNIVARIATE, Statistical Analysis System, SAS Institute, Carey, NC) evaluated the null hypothesis that the median z-score in the tinnitus group was equal to 0 and the median T-score was 50.

For the COWAT, computed median z-scores for performance reflected a comparison against age and gender vetted norms (Benton & Hamsher, 1989).

A second goal was to examine performance differences across repeated testing. A Friedman’s ANOVA assessed the differences in raw test scores across three test sessions. In post hoc analysis, Wilcoxon Signed Ranks test compared the scores in paired sessions (e.g., session 1 vs. session 2, session 1 vs. session 3, and session 2 vs. session 3). The post hoc significance alpha level was Bonferroni corrected for three tests (critical value: p ≤ .017).

RESULTS

Participant characteristics

Table 1 presents demographics and tinnitus symptoms for 14 enrolled participants. Tinnitus was bilateral in 9 and unilateral in 5 (3 rights, 2 lefts). None had hyperacusis. Hearing loss was minimal for lower frequencies (1-3 kHz). For the 8 kHz tone, hearing loss of>10 dB was generally age normal for a mean age >52 years. Pure tone thresholds (PTT) and speech reception thresholds (SRT) were binaurally similar². Audiometric examination revealed normal hearing in 4, slight loss in 5, moderate in 3, and severe in 2 participants.

Neurocognitive Test Results

For the COWAT, the median z-score and 95% Confidence Interval (95% CI) during session 1 testing was −0.26 (95 % CI: −0.65 to 0.50), which did not differ from normative data (Table 2) (Benton & Hamsher, 1989; Tombaugh, Kozak, & Rees, 1999). COWAT z-scores for sessions 2 and 3 also were normal (Table 2) and improved from test session 1 (Figure 1). However, despite the improved performance median and 95% confidence intervals of associated word numbers were similar across all three test sessions, resulting in no significant differences with repeated COWAT testing, χ²(2)=4.885, p=.72 (Figure 1A). These findings indicate no deficit in verbal fluency with repeated practice using the COWAT.

Table 2.

Tinnitus group median z-scores for number of words associated, recalled, or derived and p-values per test session for Wilcoxon Signed Rank test relative to normal population data

Test	Test Session 1		Test Session 2		Test Session 3
	Median (95% CI)*	p- value$	Median (95% CI)	p-value	Median (95% CI)	p-value
COWAT#	−0.26(−0.65 to 0.50)	0.826	0.25 (−0.41 to 0.73)	0.245	0.42 (−0.17 to 0.79)	0.220
CVLT Trial 1, List A#	−1.0 (−1.0 to −0.5)	0.004	0 (0 to 1.0)	0.094	1.0 (0 to 1.5)	0.087
CVLT Trial 5, List A#	−1.0 (−1.5 to 0)	0.067	0 (−0.5 to 1.0)	0.484	0 (−0.5 to 1.0)	0.875
CVLT Total Words Trials 1 to 5, List A&	39.0 (33.5 to 46.5)	0.005	50.0 (44.0 to 56.5)	0.967	52.5 (46.5 to 63.5)	0.318
CVLT List B#	−1.0 (−1.0 to −0.5)	0.008	0 (−0.5 to 0.5)	1.000	0 (0 to 1.0)	0.285
CVLT Short Delay, Free Recall, List A#	−0.5 (−1.0 to 0)	0.273	0 (0 to 1.0)	0.344	0.5 (−0.5 to 1.0)	0.283
CVLT Short Delayed Cued Recall, List A#	0 (−1.0 to 0.5)	0.305	0 (0 to 1.0)	0.438	1.0 (0 to 1.5)	0.152
CVLT Long Delay (20 minutes) Free Recall, List A#	−0.5 (−1.0 to 0)	0.148	0 (−0.5 to 0.5)	1.000	1.0 (0 to 1.0)	0.087
CVLT Long Delay (20 minutes) Cued Recall, List A#	−0.5 (−1.0 to 0)	0.273	0 (−0.5 to 1.0)	0.641	1.0 (0 to 1.0)	0.268
CVLT Long Delay (20 minutes) Recognition, List A#	0 (−0.5 to 0)	0.563	0 (0 to 0.5)	0.508	1.0 (0 to 1.0)	0.109
CVLT Semantic Cluster Ratio#	−1.0 (−1.5 to −1.0)	0.001	0 (−0.5 to 0.5)	1.000	1.0 (0 to 1.0)	0.268
CVLT Serial Cluster Ratio#	0 (0 to 1.0)	0.234	0 (−0.5 to 0.5)	0.844	−1.0 (−1.0 to 0)	0.022
CVLT Learning Slope#	0 (−0.5 to 0.5)	0.563	0 (−1.0 to 0.5)	0.391	−1.0	0.007
CVLT Perseveration Score#	−1.0 (−1.0 to −0.5)	0.001	−1.0 (−1.0 to −1.0)	0.0002	−1.0 (−1.5 to −0.5)	0.213
CVLT Intrusion Score#	−1.0 (−1.0 to −0.5)	0.001	−0.5 (−1.0 to 0)	0.234	−0.5 (−1.0 to 0)	0.398

^*95% CI for median was calculated in Minitab® Statistical Software

^$Wilcoxon signed rank test probability that the measured score in tinnitus was equal to a population score.

^#Normalized median z-scores in tinnitus

^&Number of recalled words in tinnitus as standardized T-scores with a mean of 50 and a standard deviation of 10

Figure 1.

Mean and 95% confidence limits for performance during three testing sessions (Session 1, Session 2, and Session 3) on standardized verbal tests. A. Controlled Oral Word Association Test (COWAT) for verbal fluency based on number of freely generated words for three specified letters. B. California Verbal Learning Test (CVLT) of number of words immediately recalled after hearing 16 words from List A during the first trial (Trial 1), from List B, and across five trials for List A. C. Number of List A words recalled after 5 trials that correctly were members of a cued semantic category following short or long (20 minutes) delays or freely recalled after a long delay. D. Derived per trial scores across 5 trials for List A. Semantic and serial cluster ratios based on the number of consecutively recalled words that were from the same semantic category or followed the original serial order of presented words. Intrusion score reflects the number of incorrectly recalled words. Inserted p-values in each plot reflect the Friedman ANOVA assessment of performance differences across the three testing sessions.

Immediate free recall of a list of heard words was below normal on session 1 testing. For example, participants immediately recalled a median of only 6 correct words for Trial 1 of List A or the one trial for List B (Figure 1B). This yielded significantly lower median z-scores (−1) for List A and List B words compared to number of recalled words in a normal population (Table 2, p=.004 and .008) (D. C. Delis, et al., 1987) Some individuals may perform poorly on Trial 1 of List A due to anxiety during the first trial for learning a word list and then recoup and perform adequately on future trials (D. C. Delis, 1989). However, factors other than anxiety possibly caused interference because the number of recalled words was still below normal on Trial 5 of List A after repeated trials of the immediate recall test. On Trial 5 a median of 11 words were correctly recalled (95% CI: 9.5 to 12.5), which yielded a normalized median z-score of −1 that trended towards significance (Table 2, 95% CI: (−1.5 to 0)). Learning rates also were significantly below normal with a median of T-score of 39 correctly recalled words across Trials 1 through 5 (Table 2, 95% CI: 33.5 to 46.5, and p=.005).

Learning and learning rates improved with repeated testing. Thus, the number of recalled words on Trial 1 for List A and the one trial for List B significantly improved from session 1 to sessions 2 and 3 (Figure 1B). For List A and B, mean number of recalled words was ~6 for session 1. During sessions 2 and 3, for List A, the mean increased to 8.3 and 9 and for List B the mean was 6.4 and 7.4 (Figure 1B). The elevated number of recalled words during sessions 2 and 3 had higher median z-scores and associated ranges that were now similar to population norms (Table 2). Learning rates also improved to respective T-scored medians of 50 and 52.5 recalled words during sessions 2 and 3, which were in the normal range (Table 2). The improved number of recalled words over that for session 1 resulted in significant results in the Friedman’s ANOVA for tests of immediate recall across the three testing intervals for List A-Trial 1, inclusive Trials 1 to 5, and List B (Figure 1B). Post hoc analyses found significantly fewer recalled words during session 1 compared to word numbers for sessions 2 or 3 and no differences between the last two sessions for each immediate recall test.

At each testing session, verbal memory for the List A words indicated normal performance at short and long delays, irrespective of free or cued recalls (Table 2). Similarly, recognition memory of learned words and distinguishing them from novel words was normal (Table 2). However, practice obtained from repeated testing resulted in significantly increased numbers of recalled words during sessions 2 and 3 for short and long delay cued recall and long delay free recall (Figure 1C).

Scores derived from patterns of word types freely recalled across the 5 trials for List A revealed qualitative deficits in encoding strategies. For example, during session 1 the Semantic Cluster Ratio indicated that only a median 1.3 sequentially recalled words (95% CI: 0.9 to 1.5) were from the same semantic category (Figure 1D). The consequent derived Semantic Cluster Ratio median z-score of −1 (95% CI: −1.5 to −1.0) was significantly below normal (Table 2, p=.001). In contrast, the median for the Serial Cluster Ratio was 2.6 recalled words (95% CI: 1.9 to 3.4) that followed the serial sequence of List A words (Figure 1D). The consequent derived Serial Cluster Ratio median z-score of 0 (95% CI: 0 to 1.0) was in the normal range (Table 2). These combined findings suggested greater reliance on learned serial word order as opposed to the more associative and efficient encoding strategy of clustering words into the same semantic category. During test sessions 2 and 3, more consecutively recalled words were from the same semantic cluster and fewer followed the original serial list order (Figure 1D). These changes suggested greater adoption of semantic encoding with practice obtained from repeated testing.

Three additional derived scores pointed towards reliance on relatively conservative recall strategies during session 1. For example, Perseveration and Intrusion median scores of 0 recalled words during session 1 (Table 2) indicated no perseverated repeats or intruded incorrect words. A finding of zero repeated and intruded words was significantly above normal (Table 2, Perseveration and Intrusion Scores with p=.0005) and might reflect close attention to the learned serial order in word List A. With a potential increased propensity for semantic encoding in subsequent testing intervals, intrusion of incorrect words increased towards normal values (Tables 2 and Figure 1D) suggestive of associative linkage in lexical memory to semantically similar words. However, perseveration scores were still above normal indicating a persistent caution against repeating already recalled words. Changes in the Learning Slope scores across test sessions were consistent with a switch in encoding strategies. During sessions 1 and 2 the median z-scores of 0 for the Learning Slope did not differ from normal, but by session 3 the median z-score of −1 was significantly below normal (Table 2, p=.007). During session 1 there was a relatively steady increase of 1.2 correctly recalled words for each successive trial with List A words. During session 3 testing, there was a drop-off to 0.55 correctly recalled words added per trial. The changes in the Learning Slope score suggest two possibilities. There might have been a ceiling effect on the usefulness of semantic encoding once it was practiced in repeated test sessions. Alternatively, patients with tinnitus had an initial cautious approach tied to the actual serial order of the learned List A and a subsequent greater reliance on semantic encoding that lead to a consequent increased number of incorrect words that also reflected the increased Intrusions Score.

DISCUSSION

Especially during the first testing session, tinnitus patients showed significant deficits in learning word lists, learning rates, immediate recall of heard words, and a serial order encoding strategy. They apparently adopted a serial as opposed to semantic encoding strategy when learning a word list for the first time. However, despite these deficits, prompted recall was normal at both short and longer delays. These findings suggest that self-reports of impaired concentration in patients with bothersome tinnitus is not a consequence of impaired memory. We discuss below several possible contributing factors to the observed cognitive deficits.

Two hypothetical factors of depression or sleep disturbances might have affected performance on the verbal tests. Nevertheless, several findings challenge this interpretation. First, tested participants had been screened using the Beck Depression Inventory-II and none had BDI scores >18. Any depression also might have persisted but did not affect improved performance during repeated testing intervals. Second, although >70% of the participants reported having sleep disturbances, prompted recall was normal even for the first testing interval. Furthermore, despite persistent sleep problems, performance improved during repeated testing intervals.

Deficits in hearing status often accompany tinnitus symptoms and this might interfere with processing spoken words. However, the current population had a variable hearing status that principally appeared to be age-related. Patients with greater deficits also did not have the lowest scores on the verbal tests. More critically, the tinnitus patients performed comparably to an age-matched normal population, especially on the COWAT verbal fluency test and CLVT recall tests. Consequently, the cognitive deficits did not occur because participants had difficulties in hearing the test words.

A more probable contribution to impaired verbal learning might be the ability to attend properly. As previously suggested, cognitive deficiencies in tinnitus possibly arise because phantom noises involuntarily garner attention resources (Møller, 2000). As previously noted, tinnitus patients perform less efficiently on tasks involving auditory verbal working memory and visual-divided attention that require voluntary, effortful control of cognitive efficiency (Rossiter, et al., 2006; Stevens, et al., 2007). In the current study, practice effects from multiple testing sessions resulted in scores that matched standardized test norms for learning and learning rate performance and a shift towards a more efficient semantic encoding strategy. These changes might indicate that after repeated test sessions the tinnitus patients possibly were more able to ignore interference from tinnitus and attend to the verbal tasks. A mechanism that might have aided the adaptation was recollecting the List A words during subsequent test sessions because memory was normal at all test intervals and even improved with practice. However, persistent verbal deficits across test sessions might have arisen because tinnitus still disrupted the associative processes involved in attending to lexical memory. For these reasons when encoding might have switched to dependence on semantic associations, errors increased that led to greater intrusion of wrongly recalled words and decreased Learning Slope scores during later test sessions. Thus, tinnitus might disturb attention to a semantic linkage when recalling words.

A prior report suggested a significant decrement in attention generally in tinnitus patients compared to a normal population based on utilization of a Stroop task (Stevens, et al., 2007). In a traditional Stroop task, attention must focus on printed words (e.g., the written word blue) and inhibit a pre-potent response to the font color of the word (e.g., shown in red). The reverse situation was also tested where the task required stating the font color for a mismatched syntax for a color name. Poor scores in tinnitus patients suggested deficits in attending to the specified category, making it more difficult to avoid distractions from a non-target competing factor. In the current study, significantly poorer scores on several tests possibly also indicated inadequate utilization of attention resources. In the CVLT, immediate recall required attending to heard words for verbal learning and attending semantic associations when encoding heard words according to specified categories within the learned lists. Thus, participants had greater difficulty with tasks involving auditory attention. It might be incorrect to explain these deficiencies from hearing loss because audiometric insufficiencies were mostly age specific. Furthermore, statistical comparisons for each test were age matched to normal populations whose hearing losses also would reflect age.

Several findings indicated that tinnitus patients might have a weakness in associative memory processes. Thus, participants did not avail themselves of a semantic clustering strategy when encoding words during the first test session. Reliance on the serial order of words in a learned list might have been less demanding because it did not require making parallel associations to semantic categories. Participants also showed high perseveration scores and low intrusion of incorrect words, which indicated caution against repeating already recalled words. A possible source of these deficits might reflect problems in sustaining or properly directing attention to lexical memory when required to make word associations or retrievals. A key aspect of retrieval is attending to memory, especially through interactions between parietal cortex attention regions and frontal cortex regions associated with retrieval (Ciaramelli, Grady, & Moscovitch, 2008). Because bothersome tinnitus especially interferes with attention, it also might disrupt the process of attending to memory and, as such, might explain a prior suggestion that people with tinnitus utilize different memory strategies compared to normal controls (Hallam, et al., 2004). However, finding no deficits in the free and cued memory tests at short or long delays or for the recognition memory test does not support suggested impaired memory after a word list was learned. Thus, the simple strategy of repetition is an important implicit aid to memory (Henson, 2003; Henson & Rugg, 2003; D.L. Schacter & Buckner, 1998; D. L. Schacter, Dobbins, & Schnyer, 2004) and thereby, might have sufficed to allow learning.

One potential cause for difficulty with verbal learning and learning rates for heard words might be interference by the tinnitus signal. Especially in participants with bothersome tinnitus, impaired cognition might reflect performing demanding tasks in the presence of involuntary attention to tinnitus, a factor that probably degrades performance. However, improved performance with repeated testing indicates normally functioning cognitive abilities, especially with practice in ignoring the tinnitus. Thus, the effects of tinnitus on cognitive efficiency are subtle, as previously suggested (Adjamian, Sereda, & Hall, 2009; Hallam, et al., 2004).

Stevens and colleagues (Stevens, et al., 2007) proposed a general depletion hypothesis as a phenomena where cognitive resources are occupied when an individual attends to their tinnitus or has thoughts related to their tinnitus. Consequently, the ability to attend to other tasks becomes a dual-task situation, which compromises performance when other demanding tasks involve controlled processes. Stevens and colleagues investigated this hypothesis by examining the performance of 11 patients with bothersome tinnitus and comparing the results to a group of controls (matched for age and IQ) on 2 visual tasks (with and without the demands of a dual task). Using the Stroop task, single baseline reaction time (RT) measurements were compared when participants had to say a read word (easy condition), say font color of viewed word (demanding condition), or perform dual tasks involving word reading or category naming while performing a concurrent RT task. Results indicated that, compared to the control group, the participants with tinnitus were slower in demanding conditions supporting the depletion of resources explanation.

Another study (Delb, Strauss, Low, Seidler, Rheinschmitt, Wobrock, & D’Amelio, 2008) in patients with severe bothersome tinnitus also supported a depletion hypothesis by demonstrating a significant centering of attention on the tinnitus signal. Delb and colleagues measured evoked potential differences to selectively attended auditory signals in controls and patients with high or low tinnitus bother. In controls and low bother tinnitus patients, an attention shift away from an auditory stimulus resulted in significant N1 amplitude changes. The high bother tinnitus group did not show a shift in the N1 response (Delb, et al., 2008). Because perceiving phantom noise depleted attention resources, these group differences suggest that tinnitus prevented shifting attention to external stimuli in the patients with bothersome tinnitus. The current results involving verbal tasks and heard words might reflect insufficient shifts in the N1 response when the examiner read the words.

The mere repetition of testing improved performance, suggesting that procedures which direct attention away from tinnitus might improve cognitive behavior (Jastreboff, 1990). One interpretation of performance improvements might reflect improved skill on test items with repeated practice. Alternatively, the repeat test session gave participants sufficient practice with performing the cognitive tasks that enough reserve remained available for normative behavior despite depleted attention resources due to tinnitus. Another possibility is that repeated neurocognitive testing provided practice that aided implicit adaptation or habituation to tinnitus.

The current findings do not distinguish between these options. However, the notable performance improvements even after repeated testing are consistent with reported reductions in experienced tinnitus by training to habituate tinnitus salience and to focus on other sensations (Searchfield, Morrison-Low, & Wise, 2007). Similarly, cognitive distraction can diminish tinnitus and lowers auditory cortex activity (Andersson, Juris, Classon, Fredrikson, & Furmark, 2006). An implied consequence of repeating the neurocognitive testing was learned skill in suppressing thoughts about tinnitus and limiting distraction from the cognitive processes needed for the tests. The most valuable conclusion from these effects of repeated testing is encouraging support that some forms of behavioral training might improve impaired concentration in people with bothersome tinnitus.

These findings might have limited applicability because the studied participants were from an extreme group of tinnitus sufferers with severe bother (i.e., Tinnitus Handicap Inventory > 38). Further research on distinguishing the specific types/domains of attention interference related to tinnitus may help to elucidate the neural pathways involved in tinnitus generated interference. It would also be useful for future research to pair neurocognitive tests with some type of objective, neurophysiological measure of attention such as evoked potentials (Delb, et al., 2008) or neuroimaging during attention demanding tasks.