Neuro-imaging studies of dual tasks generally report two types of effects. First, there is degradation of performance in one or both tasks relative to performance in either single task alone. Second, there is increased activation in the PFC in the dual task compared to each single task [5, 21, 26, 27]. Our results support the central bottleneck theory and suggest that the dlPFC is an important mediator of neural activity for both short-term storage and executive processes. To our knowledge, this is the first study to investigate specific behavioral and functional details of regional brain activation during both the dual and single components of the KPT.
In the neuropsychological field, the KPT is administered as a dual task (PR) and compared to the single tasks (P or R). The percentage of correct answers regarding reading content and picked vowels in PR decreased more than the percentage of correct story answers in R and percentage of picked vowels in P respectively. Therefore, dual task performance was diminished for both continuous and discrete task components of the KPT.
Scores obtained for picking vowels in over 20,000 people have been clearly described , but there is currently a lack of data about story substance scores. Comparison of our behavioral results of the KPT to those previously reported revealed similar vowel scores [7, 14, 28]. In our reading content results, PR reading scores were 22% lower than R scores. The variance in the reading component of R was similar to that in PR (Figure 2b). This suggests that the 22% drop in reading performance in PR was primarily due to the addition of the vowel pickout.
During these tasks, the maximal increase in BOLD signal in PFC in PR was greater than that in R. The lower reading scores in PR compared to R and greater PFC activity during PR compared to R suggests that during the PR condition there was increased dual task difficulty. Alternatively, the PFC BOLD signal cluster size in P was similar to that in PR and the differences in vowel pick-out scores were not significant. This suggests that the P task (picking the vowels out in a randomized Kana sequence) had some task complexity. We attribute this to the difficulty of picking out vowels embeded within a randomized Kana sequence with no meaning. It is easier to identify vowels within a well-known word in which one can predict the location of a given vowel.
General cortical activity
In Japanese Kana, each symbol represents a syllable or vowel, as opposed to kanji in which each character represents a word with a specific meaning. Similar to the English alphabet, Kana permits phonetic reading. Cortical activity during performance of the KPT was found bilaterally in the PFC, left SMC, bilateral PAC, and bilateral VC. Contralateral SMC in the hand region for right hand manipulation was strongly activated by mouse clicking. This activation is in agreement with SMC activity during finger tapping, finger press, and mouse clicking experiments . Increased VC activity was related to subjects viewing the screen to perform testing and similar to reports of Kana reading [30–32]. Cognitive tasks such as reading demonstrate greater activations in VC than staring at pseudo objects . However, other studies report similar VC activity for reading Kana words and non-words .
Imaging studies of Kana reading report that it is primarily processed in the left hemisphere, particularly the left dlPFC, superior and inferior parietal lobe, temporal- parietal area, and posterior inferior temporal gyrus [30–32, 34, 35]. This is similar to alphabetic reading imaging reports of neural activity in the left PFC and temporal lobe language processing areas (e.g. Broca’s and Wernicke’s areas) [30, 33, 36].
Activation of the PAC during the KPT was attributed to spatial perception for searching and retrieving vowels with voluntary control of spatial attention for both vision and touch . In particular, the superior parietal lobe has been described as a part of top-down processing and feed-forward modulation of sensory inputs, integrating visual input to execute goal-directed spatial orienting . Activity in the inferior parietal lobule, specifically BA 40, is reported during phonological processing .
In some subjects, Wernicke’s area is activated by language processing for retaining the story semantics. In young subjects, increases in Wernicke’s area (left medial temporal lobe) activity have been observed during performance of memory-encoding tasks . We suggest that the VC, PAC, and SMC activation areas are linked to the neural network associated with PFC during the KPT.
Prefrontal cortex activity
The reading task revealed relatively little PFC activity. Reading words or sentences is a well trained task and can utilize effective feedback and feedforward neural processing to follow the characters . In contrast, the vowel pick single task generated a great deal of PFC activity. Neural processing was previously reported to be greater for Kana pseudowords compared to Kana words, supporting these results . As shown in Figure 4b, some of prefrontal activation in P did not overlap that in PR. These results suggest that task P required the subjects to utilize alternate neural processing compared to the processing required for picking vowels in task PR.
Within the PFC, bilateral dlPFC activation (left BA 45, 46 and right BA 46) was related to the dual task component of the KPT . This area is highly related to verbal working memory and dual tasks [1, 41, 42]. One study found only a semantic dual-task condition activated frontal areas, including dlPFC (BA 46) . Kana and English reading experiments comparing semantic and phonological tasks reported left inferior PFC activity (e.g. BA 44, 45, 46, 47) during both types of task [20, 30, 31, 36]. This indicates that this area performs both phonological processing and semantic retrieval. The KPT required subjects to retain content in working memory while picking vowels. Memory storage in three-back tasks have demonstrated greater dlPFC compared to two-back tasks . dlPFC is reported as a brain area that mediates working memory, comprised of short-term storage and executive processes . Based on our results, we propose these components are necessary to execute the KPT.
The majority of imaging studies report primarily left hemispheric activity of the dlPFC during phonological and semantic reading tasks. Both the contrast of PR versus rest and our conjunction analyses found bilateral PFC activity during the KPT. This bilateral activity may be related to updating working memory during the continuous task (reading) of the KPT. Experiments using running span task paradigms reported bilateral activity in BA 9, 10, 11, and 46, suggesting a bilateral network was involved in memory updating . Therefore, our conjunction analyses of bilateral dlPFC may reflect more of the working memory and updating processing rather than that specific to language phonological processing. This bilateral dlPFC finding is more in line with the theory of the dlPFC as an amodal area involved both in multimodal information processing as well as storage [2–4]. The dlPFC, then, is the likely area of the functional bottleneck during the KPT dual task.
Previous investigations of dual task versus single task neural substrates have identified cortical areas that are distinct to dual tasks . Tests of the neural substrates of dual tasks suggest that dlPFC plays a crucial role in the processing required to perform these tasks but activity in this area is not implicated in single tasks (Collette et al., 2005; Jaeggi et al., 2003; Kondo, Osaka, & Osaka, 2004; Low, Leaver, Kramer, Fabiani, & Gratton, 2009; Szameitat, Schubert, Muller, & Von Cramon, 2002). Our research did not substantiate neural activity that was unique to the KPT (PR). However, identification of the dlPFC as the bottleneck region is in agreement with previous findings.
Neural imaging studies specific to the KPT are limited, with some confined to subjects with dlPFC lesions or with Alzheimer’s disease [45–47]. However, lesion and degenerative diseases cannot be construed to equate with neural activity in the healthy brain as compensatory mechanisms may be at work.
This study was only conducted on healthy young adults, therefore we cannot generalize these findings to the healthy elderly or those with MCI. This study focused on healthy young adults because PFC activity in young adults is greater during semantic cognitive tasks compared to older adults.(Logan, Sanders, Snyder, Morris, & Buckner, 2002).
Future studies will repeat this paradigm in healthy older adults and MCI patients. Further studies are also warranted to determine the sensitivity and specificity of the KPT to identify those with MCI. Previous behavioral results of single and dual task paradigms suggest total performance of elderly healthy subjects is better than those of MCI in both single and dual tasks (Montero-Odasso et al., 2009; Pettersson, Olsson, & Wahlund, 2007). This suggests that even single task components (i.e. P or R) might be able to discriminate differences in MCI versus healthy groups. However, to execute the dual task (KPT), more working memory and modified processing in dlPFC is required. Understanding the specifics of PFC activation in the dual task component of KPT (i.e. PR) substantiates that the KPT provides a more sensitive tool allowing for discrimination between groups such as healthy aging and MCI. As shown in this study, we propose BA45 and 46 are specifically being activated in healthy subjects and assume the dual task component in KPT may require processing by this area.
Activation of the dlPFC in the KPT conjuction may be due to task difficulty and not only task specificity. We did not employ an equally difficult task as a contol. However, the complexity hypothesis suggests that dual task costs (computational load) will increase as a function of task complexity. Therefore, complexity cannot be separated from the dual task.