Recent research shows that patients on sick leave report substantial problems with memory and attention (Aasvik et al., 2015). Such self-reported cognitive impairments are typically presented in relation with difficulties in social settings and everyday activities (Stenfors et al., 2013) and have been associated with work-related stress and reduced coping abilities (Broadbent et al., 1982; Folkmann et al., 1986). Sick leave is associated with a heterogeneous series of stress-related symptoms such as pain, fatigue, insomnia, depression and anxiety (Henderson et al., 2005). Patients with the aforementioned symptoms frequently report problems that go beyond the primary characteristics, such as reduced mental alertness, increased distraction, and problems with forgetfulness (Derousné et al., 1999; Mowla et al., 2008; Kronholm et al., 2009; Tesio et al., 2015). A wealth of studies have also found evidence of cognitive dysfunctions, the most consistent findings being related to impairments in working memory (WM) and executive functioning (Christopher and MacDonald, 2005; Derakshan et al., 2009; Constant et al., 2011; Fortier-Brochu et al., 2012; Landrø et al., 2013). Despite the impact and frequency of cognitive deficiencies, clinicians have primarily focused on treating the overarching symptoms i.e., the insomnia, the pain, the fatigue and so forth, with less focus given to treating the cognitive impairments.

WM has been implicated in a wide variety of complex cognitive and behavioral processes including selective attention, memory and learning, emotional regulation, and pain perception (Ochsner and Gross, 2005; Oosterman et al., 2010). Even very small differences in capacity may have huge consequences in various domains including academic achievement, occupational functioning and emotional health (Alloway and Alloway, 2010; Joormann and D'Avanzato, 2010; Oosterman et al., 2010; Alloway et al., 2013). Hence, examining if focused WM training improves cognitive function and perceptions of memory functioning is highly relevant in the rehabilitation of cognitive deficits.

WM is considered a mental workspace, and may be defined as the ability to actively hold and manipulate information in mind, while simultaneously filtering out or inhibiting distractions from entering the active state (Kane and Engle, 2000). Inhibitory control refers to the ability to resist or filter out distraction, and cancel irrelevant or no longer relevant responses (Miyake et al., 2000). According to the aforementioned definition of WM, inhibitory control is essential in guarding information processing in WM, protecting processing from irrelevant or distractive stimuli. Apart from the role inhibitory control plays in WM, it is also recognized as an underlying mechanism involved in almost all complex cognitive and behavioral processes.

When it comes to WM training, the main question is if effects of WM training extend beyond the specific tasks presented within the training program. Transfer of training effects is a debated topic, and although there are some promising results (Klingberg et al., 2002, 2005; Westerberg et al., 2007; Jaeggi et al., 2008; Brehmer et al., 2012; Salminen et al., 2012; Anguera et al., 2013; Au et al., 2015; Waris et al., 2015), some also report negative findings (Owen et al., 2010; Redick et al., 2012; Zinke et al., 2012). For reviews see (Klingberg, 2010; Shipstead et al., 2012; Melby-Lervåg and Hulme, 2013). The inconsistent results highlight the need to identify factors that may moderate transfer, and the nature of the underlying mechanisms through which transfer may occur. Because inhibitory control makes such an intrinsic and integral part of working memory, and because it also has the potential to act on a broader landscape of cognitive abilities, it may be defined as one of the unique mechanisms that promote transfer from WM training. In addition, an important but often somewhat neglected aspect of WM training is whether such training has the ability to improve subjective perceptions of cognitive functioning. Such self-perceptions may be just as relevant when considering the practical value of this type of intervention. Following this, an essential question is if WM training is better or more efficient in targeting this aspect than other cognitive treatment programs.

The current study aimed to examine if adaptive WM training improves objective cognitive functioning by strengthening performance on a test of spatial working memory and a test of inhibitory control. We also sought to examine if adaptive WM training improves subjective cognitive functioning by improving perceptions of everyday memory functioning. Our first hypothesis was that participants receiving adaptive WM training would perform better on post-tests of spatial working memory than the comparison group. Our second hypothesis was that participants who received adaptive WM training would demonstrate better response inhibition compared to the comparison group as demonstrated by significantly lower Stop Signal Reaction Time (SSRT). Our third hypothesis was that participants receiving adaptive WM training would report substantially less subjective memory complaints at post-testing than the comparison group.

The study was approved by the Regional Committee for Medical and Health Research Ethics (2013/634 REK-midt) and adhered to the Helsinki Declaration. All participants were given a complete description of the study and gave written informed consent before inclusion.

Descriptive data including self-reported levels of pain, insomnia, fatigue, depression, anxiety and everyday memory failures are presented in Table 2. The experimental group consisted of 25 subjects (21 females and 4 males) and the control group consisted of 29 subjects (23 females and 6 males). There was some difference in age between the two groups. Analyses confirmed that the difference in age was not statistically significant (p-value = 0.065). The comparison group had been on sick leave for an average of 15 months (M = 15 months), while the training group had been of on sick leave for an average of 14 months (M = 14 months). The groups were similar in general cognitive ability; they reported similar levels of pain, insomnia, fatigue, depression, and anxiety, all within what is considered subclinical range. Both groups reported considerable memory problems. Their scores were higher than what has been found in MS and stroke patients (Royle and Lincoln, 2008).

Pre/post means, standard deviations and paired samples t-tests of main outcome variables are presented in Table 3. The results indicated a substantial time reduction in SSRT-last half in the experimental condition (p-value = 0.033) but not in the control condition (p-value = 0.973). The correlated t-tests indicated no statistically significant changes in SWM strategy from pre to post-tests in either of the groups. Both groups demonstrated significantly less memory problems at post-testing.

As presented in Table 4A, the ANCOVA showed no statistically significant difference between the two groups on the post-test measure of SWM Strategy. Hence, the Cogmed training did not translate into better performance on SWM post-test relative to the comparison group. Neither was there an effect of age, gender or group by gender. As presented in Table 4B, the ANCOVA showed a significant training effect on post SSRT-last half. The subjects in the training condition performed considerably better than the subjects in the control condition on post-tests of inhibitory control (p-value = 0.025). The effect measure shows an improvement among the subjects in the training condition relative to the subjects in the control condition. Pre-test scores on SSRT-last half and age co-varied with post-test SSRT scores. There was an effect of age, but there was no effect of gender, nor was there a group by gender effect. We did a Spearman Correlation analysis to examine if the improvement in WM after Cogmed training (index improvement, see Table 1) was associated with the change/improvement in the post-test of inhibitory control (SSRT), we found a significant correlation (0.386), p-value 0.035 between the two measures. As presented in Table 4C, the ANCOVA showed no significant group effects on post scores on the EMQR. In addition, there was no effect of age, gender, or group by gender.

The present study examined the effects of a 5 week adaptive WM training program (Cogmed) in a group of patients currently on sick leave due to symptoms of pain, insomnia, fatigue depression and anxiety. Training effects were assessed by comparing performance on three tasks assessing spatial working memory, inhibitory control (Stop Signal Reaction Time), and perceptions of everyday memory functioning. We found no evidence in support of our first hypothesis, stating that participants in the training condition would perform better than the comparison group on the post-test of spatial working memory. We did, however, find support for our second hypothesis, stating that participants who received WM training would perform better than the comparison group on post-tests of inhibitory control. Moreover, we found no support for our third hypothesis stating that participants who received WM training would report substantially less memory problems compared to participants in the comparison group.

While both the Cogmed program and SWM test target spatial working memory, they do it in slightly different ways. Although somewhat speculative at this stage, it may be that the somewhat different test designs may help explain the discrepancy in results with regard to the test of SWM and the test of inhibitory control. The Cogmed program consists of several spatial working memory tasks with different designs. One of the tasks, named “chaos” involves keeping track of an increasing number of specific objects presented within a larger matrix of objects. All of the objects move and continuously change positions. The program signals the objects that have to be remembered by briefly changing their color. Because only a few of the objects have to be remembered, the remaining objects act like distractors which must be blocked or inhibited to keep track of the selected objects. The “chaos” task therefore targets both SWM and the ability to block or inhibit distractors. As the number of items that has to be remembered increases, the task requires more effortful control of attention thereby putting greater demands on inhibitory control. The SWM test involves keeping track of a single object, which shifts position among an increasing number of stationary objects. Because the objects do not move, they might not be as intrusive or potent distractors as if they had moved. Thus, while the SWM test measures the ability to maintain spatial information in WM, it does not make the same requirements for inhibitory control and therefore to a lesser degree, measures the ability to inhibit distractive stimuli. Hence, while the chaos task may have trained inhibitory control, the improved ability to inhibit distractors is not as targeted within the SWM test paradigm, and this may explain why the improved performance in inhibitory control did not translate into better performance on the SWM task. Our results may suggest that inhibitory control is more accessible and susceptible to transfer of training effects than SWM per se, future research should examine if this is indeed the case.

Because inhibition in and by itself has broad implications across various cognitive and behavioral processes including perceptual and emotional regulation, this result warrants attention (Ochsner and Gross, 2005; Joormann, 2010; Oosterman et al., 2010; Miyake and Friedman, 2012). There was a difference between the groups in regard to the amount of treatment that they received. While the training group received WM training in addition to the rehabilitation program, the control group only received the rehabilitation intervention. As a result we cannot exclude the possibility that the difference in treatment intensity might have had some effect in terms of the improved inhibitory control. However, because we found a significant correlation (p-value = 0.035) between the Cogmed Index Improvement and the improvement in inhibitory control (SSRTs) it indicates that the WM training was indeed related to the improvement in inhibitory control. Still, the present finding should be considered as a preliminary result and must be replicated by other studies. When it comes to the improvement in inhibitory control (SSRT), it might seem modest. however, it might be relevant in situations that require a high level of attentional focus and rapid decision-making in which the ability to quickly inhibit or filter out distracting information is crucial such as in a complex work environment, or in terms of strengthening the ability to regulate and control impulsive and/or compulsive behaviors (Dalley et al., 2011).

The negative finding with regard to perceptions of everyday memory functioning might reflect that the clinical intervention offered to both groups had a strong effect on this specific measure. Alternatively, it might reflect a “feel good” or an expectancy effect related to the interventions. Some studies have reported that subjective memory problems are associated with objective cognitive impairments (Grace et al., 1999; Landrø et al., 2013; Tesio et al., 2015). It seems reasonable to assume that the improvements in inhibitory control following training (such as in our study) would also generate a greater reduction in self-reported memory problems, but this was not the case. A number of studies have failed to find any association between self-reported and objective measures of memory impairments (Suhr, 2003; Glass et al., 2005). The lack of association may reflect that self-report questionnaires and neuropsychological tests relate to different concepts, tapping into different sources and facets of information (Williams et al., 2011). While self-report questionnaires may reflect metacognitive beliefs and metacognitive monitoring (Jacobsen et al., 2016), such subjective information typically lies beyond the scope of neuropsychological tests, such as the SST test.

Several studies have examined effects of WM training on measures of inhibition (Klingberg et al., 2002, 2005; Olesen et al., 2004; Westerberg et al., 2007; Thorell et al., 2009; Chein and Morrison, 2010; Van der Molen et al., 2010; Dahlin, 2011; Anguera et al., 2013; Berkman et al., 2014). This specific field is characterized by extensive cross-study variability; focusing on different subgroups (i.e., children and young adults with and without ADHD, students, elderly, healthy, stroke patients, etc.), using different tests to measure inhibition (Stroop, go/no go tests etc), and finally using different training protocols (Cogmed, OddYellow, Neuro Racer etc.). All of the aforementioned factors complicate and confuse attempts to generalize results. The mixed findings underscore the fact that training effects may vary due to specific characteristics such as age, neurological, psychological and cognitive factors. As mentioned above, studies differ in regards to which tests they have used to tap inhibition. The Stroop task has been used in several studies (Klingberg et al., 2002, 2005; Olesen et al., 2004; Westerberg et al., 2007; Thorell et al., 2009), but due to flaws in the administration of the test (not including congruent trials) some of those studies (Klingberg et al., 2002, 2005; Olesen et al., 2004) have been criticized because performance cannot be interpreted as related to inhibitory control (Shipstead et al., 2012). Still, some studies have found promising results in terms of improved inhibitory control (Chein and Morrison, 2010; Anguera et al., 2013; Berkman et al., 2014). Because training protocols vary in regards to task design, the number of tasks, and the intensity and length of training, results stemming from one training program may not be readily replicated by studies using different training protocols (Morrison and Chein, 2011). The mixed results may reflect that different WM training programs (more or less effectively) target different mechanisms in different subgroups depending on the neuropsychological profile of each group.

Recent theories and research suggest that deficiencies in inhibition may contribute to the development and maintenance of symptoms such as pain, insomnia, fatigue, depression and anxiety (Ochsner and Gross, 2005; Derakshan et al., 2009; Joormann, 2010; Joormann and D'Avanzato, 2010; Miyake and Friedman, 2012). In a fairly recent study researchers were able to demonstrate a direct link between pain sensation and cognitive inhibition; better inhibitory control was associated with lower pain sensitivity (Oosterman et al., 2010). Our results indicate that inhibition may be modulated and improved by adaptive WM training. Targeting inhibitory control may thus offer considerable promise in terms of improving the effectiveness of traditional therapeutic interventions (Siegle et al., 2007).

Our sample included a series of stress related and comorbid symptoms, all of which have been associated with impairments in WM. Although the WM profile might be similar across symptoms, there might be differences in terms of how the different symptoms affect the ability to profit from the WM training. The effect of these symptoms with regard to transfer effects could not be examined due to lack of statistical power. Future studies should address this aspect as it might be an important factor that moderates the potential to benefit from such training. Moreover, the duration and sensitization of different pain and fatigue states may not be linked to the duration of sick leave, making it difficult to sample a homogenous group of participants. We did not include tests to assess verbal-numeric WM although this was targeted by the training program. This study is limited by the fact that we did not include additional transfer measures for each of the cognitive domains that we examined. Furthermore, because the post-tests were conducted within 3 weeks of completing the training program the current results therefore reflect short-term effects. While this study shows an improvement in inhibitory control following training, this is only at test level and it may not transfer to real life.

WM training was associated with improved inhibitory control as indicated by a statistically significant reduction in SSRTs among subjects in the training condition compared to the control condition. WM training did not improve performance on the SWM transfer measure. Self-perceived memory functioning did not differ between groups at post-testing. Improving inhibitory control may support working memory by strengthening the ability to quickly suppress or block distractive stimuli.

All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.