Introduction

Evolving from early classic paper-and-pencil tests, neuropsychological assessment has been modernized over the past decades with advances in cognitive neuroscience and information technology (Marcopulos & Lojek, 2019). To facilitate precision neuropsychology and overcome the shortcomings of traditional paper-and-pencil tests, such as over broadness and lack of process purity, a number of cognitive neuroscience paradigms have been developed for specific cognitive processes (Kessels, 2019). Tests and questionnaires were gradually computer-administered, making instructions more standardized and measurement of speed and responses more consistent (Howieson, 2019). More recently, gamification has become another modern trend in neuropsychological assessment, driven by the need to increase productivity and efficiency in fun ways (Koivisto & Hamari, 2019).

Despite rapid progress in the field of psychological assessment, the assessment of its psychometric properties, mainly concerned with reliability and validity (Asunta et al., 2019), has not progressed at a comparable pace. Most studies reporting validity data of psychological tests were based on classic psychometrics, such as patterns of convergence and divergence by correlating tests with other measurement tools (Pawlowski et al., 2013). Classic psychometrics has a scientific basis and advantages, especially when examining the validity of traditional paper-and-pencil tests with fixed outcome measures. However, modern psychological assessments often have multiple outcome variables, such as reaction time (RT), accuracy, error rate, and composite/ratio scores. Different measures of the test can lead to different validation results. Webb and Demeyere (2022) investigated the convergent validity of an executive function task, through 10 different outcome variables divided into four categories: time, accuracy, error, and composite/ratio. They found significant differences in correlation coefficients when assessing convergent validity through different outcome variables. While some lead to small and insignificant correlations, others lead to larger significant correlations (Webb & Demeyere, 2022). When multiple outcome variable assessments are available, it is difficult to determine the true validity and which variable best reflects the targeted cognitive process.

To solve the current psychometric dilemma of psychological assessments, validity assessment should also be modernized (Bilder & Reise, 2019). Incorporating cognitive neuroscience and brain imaging techniques to validate assessments through the observation of brain activity can provide a more reliable approach to establish its psychometric properties. Unlike classic psychometrics, which indirectly examine validity based on correlations between variables (Borsboom et al., 2004), brain imaging validation avoids multiple confounding outcome variables and directly assesses targeted cognitive processes. If a task is developed effectively, key brain regions responsible for the targeted cognitive process should be activated when participants perform the task. For example, the dorsolateral prefrontal cortex consistently exhibits activation during working memory tasks and is considered a crucial region (Evangelista et al., 2021; Toepper et al., 2010). The inferior frontal gyrus (IFG), particularly the right IFG, has been identified as a key region involved in inhibition control (Aron et al., 2004; Puiu et al., 2020; Xue et al., 2008).

The degree of activation of key brain regions should be significantly correlated with task performance (Jia et al., 2020a). Following these principles will help determine the best representative measure of a task among multiple outcome variables, i.e., variables closely related to brain activity. In addition to classic psychometrics, brain imaging validation can serve as an important approach to ensure effective and comprehensive assessment of newly developed cognitive tasks. However, the directions of the relationships between the brain activation and task performance are not always consistent. Both positive and negative relationships were revealed in previous literature (Meinzer et al., 2012; Suda et al., 2020; Zhang et al., 2013). According to the compensation-related utilization of neural circuits hypothesis (CRUNCH) (Reuter-Lorenz & Cappell, 2008), the direction of the relationship between brain activation and task performance depends on task difficulty. In simple tasks, the brain requires less activation to achieve optimal performance because it efficiently utilizes minimal effort to accomplish the task. Conversely, in more challenging tasks, a positive relationship is observed, indicating that increased brain activation and effort are needed to support better performance.

Agile Fisherman is a cognitive flexibility task that we recently developed and has been shown to be reliable and valid through classic psychometrics (Wang et al., 2022). Cognitive flexibility is the ability to adjust thinking or approaches to changing situations, demands, or priorities. It involves the ability to switch between different mental tasks or strategies, to switch perspectives, to generate alternative solutions, and to adapt to new information or circumstances (Diamond, 2013). The inferior frontal junction (IFJ) has consistently been reported to play a key role in cognitive flexibility (Derrfuss et al., 2005; Kim et al., 2012; Worringer et al., 2019). The degree of IFJ activation is different in young and older adults. Older women showed greater activation of the left IFJ when performing a switching task compared to young women (Kuptsova et al., 2016). A meta-analytic study shows that the brain region with significant differences in activation between young and older adults during cognitive flexibility tasks is located in the IFJ (Heckner et al., 2021). In addition to the IFJ, age-related differences in brain activity related to cognitive flexibility have also been observed in other regions. When performing a number switch task, Nashiro et al. (2018) found that while the right posterior cingulate cortex and right superior frontal gyrus activated greater in older adults relative to young adults, the left insula and bilateral frontal gyri in young adults activated greater than in older adults. In a letter-switching task, older adults showed significantly greater activation than young adults in a wide range of fronto-temporal regions, including the dorsolateral prefrontal cortex, pars operculum, superior temporal gyrus, and posterior and anterior cingulate gyrus (Eich et al., 2023).

Recently, some researchers have adopted state-of-the-art methods to study the brain mechanisms underlying cognitive flexibility. Capouskova et al. (2023) used deep autoencoders to uncover the brain’ two-dimensional integrated and segregated states manifolds. They revealed that the brain achieves cognitive flexibility by flexibly switching between transient integrated and segregated functional connectivity (FC) states. In a previous study, Cabral et al. (2017) showed a strong correlation between the switching dynamics of FC at rest and cognitive performance in older adults. Training artificial neural networks through deep learning provides another potential approach to investigate the brain mechanisms underlying cognitive flexibility (Yang et al., 2019a). By training a recurrent artificial neural network to perform 20 interconnected cognitive tasks, the study revealed that the compositional neural representation of rules acquired during the learning process was a crucial feature of cognitive flexibility, that is, one task can be performed by recombination instructions from other tasks (Yang et al., 2019b).

In the current study, we aimed to further examine the validity of Agile Fisherman through traditional brain imaging in both young (aged 18–35) and older adults (aged 60–80). Specifically, the univariate functional activity during the game was initially performed to determine whether key brain regions were activated and to compare the age-related differences in activation in these regions. Subsequently, the neural representations and whole-brain FC of the key regions were analyzed using multivoxel pattern analysis (MVPA) and psychophysiological interaction (PPI) analysis, respectively, to further explore the underlying brain mechanisms behind the functional activity of cognitive flexibility. We hypothesize that if Agile Fisherman is an effective cognitive flexibility task, the IFJ will be activated when participants perform the task, and the behavioral performance will be significantly correlated with the degree of IFJ activation. Furthermore, we hypothesize that if the task is effective, it will be age-sensitive, with young and older adults having varying degrees of IFJ activation when performing the task.

Methods

Participants

The literature has shown that young adults, aged between 18 and 35 years old, typically exhibit stable cognitive and brain function (Lövdén et al., 2010). In contrast, older adults, aged over 60 years old, experience noticeable declines in cognitive and brain function, while middle-aged adults are in a transitional phase from stable to declining (Grady, 2012; Park & Festini, 2016). Consequently, researchers frequently utilize young and older adult to investigate age-related differences in cognitive and brain activity, aiming to investigate the underlying mechanisms of aging (Baez-Lugo et al., 2023; Cutler et al., 2021; Kim & Kim, 2022; Zhang et al., 2023). Therefore, the current study also included young and older adults to examine age-related differences in brain activity during the execution of the cognitive flexibility game Agile Fisherman.

We recruited 90 participants through online advertisements. Participants were included if they were aged 60–80 years for older adults and 18–35 years for young adults, had at least 6 years of education, had normal or corrected-to-normal vision, had a Mini-Mental State Examination (MMSE) score of at least 24, had no severe physical or psychiatric disorders. After screening, 69 eligible participants were included, and 7 participants were excluded due to excessive head motion. Ultimately, 62 participants (31 older adults and 31 young adults) were included in the analysis. The two groups had similar gender ratios and years of education, but young adults had significantly higher MMSE scores than older adults. Demographic information for the participants in each group is listed in Table 1.

Table 1 Demographic information and behavioral performance of both groups of participants
Full size table

Gamified cognitive flexibility task

Agile Fisherman is a fishing game for cognitive flexibility assessment with high reliability and validity at the behavioral level (Wang et al., 2022). The game design is based on the color/shape cued switching task paradigm (Miyake & Friedman, 2012), a well-established and widely recognized paradigm for assessing cognitive flexibility. In the gamified task, a fishing net appears in the middle of the sea world, and a fish and a shrimp randomly appear in eight white bubbles around the fishing net. The game includes two types of fishing nets, with or without handles, representing two different tasks. If the net has a handle, participants should catch marine life according to the color of the net (red or blue); if the net does not have a handle, participants should capture marine life based on the shape of the net (fish-shaped or shrimp-shaped). Task types are divided into repeat conditions and switch conditions. The former indicates that the current task requirements are similar to the previous one, and the latter indicates that the current task requirements are different from the previous one. Under the repeat conditions, participants receive 5 points if they catch marine life correctly; otherwise, they will lose 5 points. Under the switch conditions, participants score 10 points for correctly capturing marine life and lose 10 points for incorrectly capturing marine life. Measurements of behavioral performance included final game score, repeat RT, switch RT, switch cost (switch RT – repeat RT), and composite Z score (sum of transformed negative Z of repeat RT and switch RT as well as transformed Z of the accuracy).

Considering the compatibility and operability of Agile Fisherman during MRI scanning, the current study adapted it into an fMRI version by employing an event-related design. A total of 160 trials were included in the task for two runs, each of which included 80 trials and lasted 8 min and 10 s. The details of the task are shown in Fig. 1. Specifically, the net and marine life appear simultaneously for 1000 ms, and participants should respond depending on the type of net within 2000 ms. The trials were presented in pseudorandom order, with a 1:1 ratio of repeat and switch trials. The trial interval varied from 2000 ms to 4000 ms. The task was operated through a custom-designed fMRI-compatible response pad with eight buttons, each corresponding to a bubble in the same direction. Before the formal scan, participants practiced through a simulated MRI scanner to familiarize themselves with the response pad and the task. When the accuracy reached 70%, a formal scan was conducted.

Image acquisition and preprocessing

The experiment was conducted on a 3T MRI scanner (GE Discovery MR750) in the IPCAS. Foam padding was used to support the head and neck to minimize head motion and reduce cumulative head drift during scanning. Functional T2*-weighted images were acquired using gradient-echo echo-planar imaging (EPI) sequences with a top-down sequential order during the task according to the following parameters: repetition time (TR) = 2000 ms, echo time (TE) = 30 ms, flip angle (FA) = 90°, voxel size = 3.5 × 3.5 × 3.5 mm3, slice thickness = 3.5 mm, slice number = 37 slices, matrix = 64 × 64, and field of view (FOV) = 224 × 224 mm2. The fMRI data were preprocessed through the SPM12 software (https://www.fil.ion.ucl.ac.uk/spm/). Preprocessing steps included head motion correction, slice timing correction, normalization, and spatial smoothing with a Gaussian kernel of 7 mm full width at half maximum (FWHM). Participants with large head motions (≥ 3 mm) were excluded from further analyses. The automated anatomical labeling (AAL) atlas (Tzourio-Mazoyer et al., 2002) was used in the following analyses for identification of brain regions.

Univariant functional activity analysis

After preprocessing, the 1st level analysis was conducted to assess the blood oxygen level-dependent (BOLD) signal under repeat and switch conditions through the general linear model (GLM). Then, the model parameters were estimated, and single-participant statistical contrast maps of the switch condition > repeat condition were constructed. In the 2nd level group analysis, a one-sample t test was applied to identify brain regions activated when performing the task in older and young adults. Whole brain results were corrected for multiple comparisons using a familywise error (FWE) threshold of p < 0.05.

To investigate the relationship between the activation of significant brain regions and game performance, the average beta values of these regions were extracted, and Pearson correlation analyses were conducted between brain activation and game performance in older adults and young adults.

Regions exhibiting significant correlations with game performance were considered regions of interest (ROIs). To investigate the age sensitivity of the Agile Fisherman at the neural level, we compared the BOLD signal intensity and percent signal change (PSC) within ROIs between older and young adults. BOLD signal intensity was the average beta value extracted from the statistical contrast maps, representing the relative brain activity of the switch condition versus the repeat condition in the current study. PSC refers to the difference in fMRI signal between the baseline condition (B) and the task condition (T), calculated as (T-B)/B×100%) (Jia et al., 2020b). In this study, the PSC within ROIs under the repeat condition and switch condition was calculated using the Marsbar toolbox (https://marsbar-toolbox.github.io/index.html). A two-sample independent t test was conducted to compare BOLD signal intensity and PSC within the ROIs between older and young adults.

Multivoxel pattern analysis

We further assessed the decoding proficiency of older and young adults in representing different game conditions via MVPA, which can detect fine-scale spatial patterns. First, for the 1st level analysis, we performed the single-subject GLM procedure to separately estimate brain maps of the three game conditions (switch, repeat and baseline). Then, we conducted ROI-level MVPA to decode two task conditions using the Nilearn package (Abraham et al., 2014) in all the participants. The classifiers were trained using linear support vector machines with default parameter. The leave-one-out cross-validation was used to evaluate classification performance of the switch and repeat conditions compared to the baseline. Classification performance is assessed based on mean classification accuracy. To improve the robustness of the results, each cross-validation was repeated ten times (Valente et al., 2021). We then implemented permutation tests of 1,000 task label randomizations and replicated MVPA pipeline to determine whether the accuracy was significantly higher than the chance level (He et al., 2021). Significance was statistically inferred from a null distribution of accuracy and was calculated by yielding permutations that is larger than the actual value.

Psychophysiological interaction

PPI analyses (Friston et al., 1997) were performed through SPM12 to measure the FC of volume of interest (VOI) with other brain areas in time series. The ROIs selected in the univariant functional activity analysis were chosen as the VOIs. For each participant, the time series of VOIs were extracted from the 1st level analysis of univariant functional activity analysis with an uncorrected threshold of p < 0.05. Then, the subject-level GLM model for PPI analysis was constructed, requiring three main regressors: physiological variable, psychological variable, and PPI variable. The PPI variable was calculated as the element product of the deconvolved VOI time series (physiological variable) and a vector coding for the effect of the task (switch > repeat, psychological variable). These three regressors were re-convolved with the hemodynamic response function (HRF). The images of the interaction regressor from all participants were loaded into subsequent 2nd level PPI analyses for one-sample t test in older and young adults. Multiple comparison corrections were performed for whole brain results using the FWE threshold of p < 0.05.

To investigate the relationship between PPI and game performance, the PPI values between VOIs and significant brain regions were extracted. Pearson correlation analyses were conducted between PPI values and game performance in older adults and young adults, respectively.

Results

Behavioral performance

The game score and behavioral performance on the gamified cognitive flexibility task in older and young adults are presented in Table 1. The independent sample t test showed that, except for switch cost, young adults performed significantly better than older adults in game score (t = -5.08, p < 0.001), repeat RT (t = 5.78, p < 0.001), switch RT (t = 5.91, p < 0.001), accuracy (t = -5.08, p < 0.001), and composite Z score (t = -7.16, p < 0.001) (Table 1, Figure S1).

Univariant functional activity

Whole-brain activation analyses found that the left IFJ, inferior parietal lobule (IPL), and superior parietal lobule (SPL) were significantly activated in older adults (Table 2). The left IFJ activation intensity was significantly correlated with the composite Z score (r = 0.37, p = 0.042) (Fig. 2a). No other significant correlations were found.

Table 2 Clusters activated in switch-repeat condition
Full size table

For young adults, the left middle frontal gyrus (MFG), IFJ, and inferior frontal gyrus (IFG) were significantly activated during the task (Table 2). Left IFJ activation intensity was significantly correlated with the composite Z score (r = 0.37, p = 0.039) (Fig. 2b). No other significant correlations were found.

Considering that the left IFJ cluster activated in young adults was included in the cluster activated in older adults, the latter was selected as the ROI. The results revealed no significant difference in BOLD signal activation intensity in the left IFJ between older and young adults (t = 0.15, p = 0.879). However, under both the repeat and switch conditions, the PSC of the left IFJ in older adults was significantly higher than that in young adults (t = 2.97, p = 0.004; t = 3.06, p = 0.003) (Fig. S2).

Multivoxel pattern analysis

The left IFJ cluster activated in older adults was selected as ROI in the MVPA. Within the ROI, the classification accuracy of the switch condition was 0.85 and 0.82 in older and young adults, respectively. Similarly, the classification accuracy of the repeat condition was 0.85 and 0.81 in older and young adults, respectively. All values are significantly higher than the random level, and older adults exhibited higher accuracy than young adults, indicating that older adults could identify switch condition and repeat condition more precisely than young adults.

Psychophysiological interaction

Consistent with above analyses, the left IFJ cluster activated in older adults was chosen as VOI in the PPI analysis. One-sample t test revealed the left IFJ exhibited a significant FC with the left IPL and the right precuneus during the cognitive flexibility task in older adults (Fig. 3a). In young adults, the left IFJ demonstrated a significant FC with the left MFG and the right angular during the game (Fig. 3b). Furthermore, in older adults, FC of the left IFJ with the both left IPL and the right precuneus was negatively correlated with the game performance (r = -0.45, p = 0.010; r = -0.38, p = 0.036) (Fig. 3c), suggesting that stronger FC was associated with poorer task performance in older adults.

Discussion

The current study examined the validity of the gamified cognitive flexibility task at the neural level. Consistent with our hypothesis, the left IFJ was activated in both older and young adults when performing the task. Located at the junction of the inferior frontal sulcus and the inferior precentral sulcus, the IFJ is consistently found to be activated in varying types of cognitive flexibility paradigms and is considered a core region of cognitive flexibility (Kim et al., 2012; Worringer et al., 2019). Brain activation observed in the cognitive flexibility task is consistently and reliably located on the left IFJ, although stereotaxic coordinates are variable (Derrfuss et al., 2009).

The present study revealed significant positive correlations between IFJ activation and behavioral performance in older and young adults. Consistent with previous findings (Armbruster-Genc et al., 2016; Yin et al., 2018), these results suggest that stronger IFJ activation is associated with better cognitive flexibility performance, further reflecting the role of the IFJ in cognitive flexibility. It should be noted that IFJ activation is correlated with the composite Z score, a comprehensive performance measure of RT and accuracy. Combining accuracy and RT provides a comprehensive picture of overall performance and implicitly controls for any variation in speed-accuracy trade-offs (Lyons et al., 2014). Therefore, the composite Z score is the most representative index among the outcome variables of Agile Fisherman. For the future use of this cognitive flexibility assessment tool, the composite Z score is recommended as the primary measure, and other outcome variables can be considered secondary and supplementary indices.

Although there was no significant difference in activation intensity in the left IFJ between older and young adults, PSCs in older adults were significantly higher than those in young adults under both switch and repeat conditions. A previous study demonstrated that contrast measurement might reduce the signal-to-noise ratio and individual differences (Hedge et al., 2018), which could mask differences between groups and explain the lack of differences in activation intensity between older and young adults in our study. PSC, with no need for a control condition, was suggested to be more understandable and meaningful than activation intensity values obtained through comparison calculations (Yuan et al., 2023). The significant group differences found in PSC demonstrated the age sensitivity of the Agile Fisherman test. Despite showing higher PSCs in both the repeat and switch conditions, older adults performed worse than young adults, which might reflect a decrease in neural efficiency as they age, requiring greater activation for the same task (Gregory, 2020; Morcom & Henson, 2018; Nyberg et al., 2012). According to Park and Festini (2017), the neural inefficiency in older adults was due to a lack of cognitive resources. Tagliabue and Mazza (2021) attributed the neural inefficiency to changes in cognitive resource allocation, suggesting that older adults were more attracted to distractors and tended to allocate limited cognitive resources to distractors, leaving fewer resources for targets.

MVPA demonstrated that older adults exhibited more precise representations of both the switch and repeat conditions compared to young adults. We speculated that the higher PSCs observed in older adults contributed to their enhanced representation of different conditions in the gamified cognitive flexibility task. Du and Zatorre (2017) found that brain regions displaying stronger activation in a syllables task also exhibited greater specificity in phoneme representations. However, the specific relationship between the intensity of brain activity or PSC and the precision of neural representation requires further investigation. Notably, the age differences in representation precision between the switch and repeat conditions were similar, and the age differences were diluted when comparing the difference of switch versus repeat condition. The current results further highlight that contrast measurements can mask differences between groups.

In addition to the left IFJ, the left IFG and MFG were also found to be significantly activated when young adults performed the gamified cognitive flexibility task, consistent with previous findings (Kim et al., 2012; Worringer et al., 2019). Cognitive flexibility is not a simple mental process. It requires and builds on inhibition to deactivate previous views or rules and utilize working memory to activate different views or rules (Diamond, 2013). IFG is believed to facilitate inhibition and appropriate behavioral control. IFG activation in young adults highlights the intrinsic role of inhibition in purposeful flexibly adjusted behavior in cognitive flexibility (Zuhlsdorff et al., 2022). The activation of the MFG in young adults may be due to the emphasis on maintaining and manipulating task rules in working memory while performing the gamified cognitive flexibility task (Kim et al., 2012). Previous results suggest that the IFG and MFG interact with the IFJ to mediate response shifting in cognitive flexibility (Zuhlsdorff et al., 2022). This is consistent with results of PPI analysis in the current study, revealing a significant FC between the left IFJ and left MFG in young adults. These findings suggest a collaborative role of the left IFJ and left MFG in facilitating cognitive flexibility in young adults.

In older adults, the left IPL and SPL were activated while performing the gamified cognitive flexibility task. These regions, situated in the posterior parietal cortex, have been implicated in interrupting current cognitive activity and reorienting attention to another cognitive activity during cognitive flexibility tasks (Corbetta et al., 2008; Ptak, 2012). While older adults tend to activate parietal regions in addition to prefrontal regions, young adults typically recruit more prefrontal regions and deactivate parietal regions during cognitive flexibility tasks (Eich et al., 2016). These differences may reflect age-related brain function changes and reconfigurations (Kunimi et al., 2016). The observed significant activation of the left IPL was in line with the results of the PPI analysis, which demonstrated a significant PPI between the left IFJ and left IPL in older adults. These findings suggested that the left IFJ and left IPL worked together to support cognitive flexibility in older adults. Additionally, the PPI analysis revealed that older adults exhibited worse task performance when the left IFJ showed stronger FC with the left IPL and the right precuneus. This phenomenon might be interpreted in the context of brain dedifferentiation during the aging process. Dedifferentiation refers to a decrease in neural activation specificity and an increase in neural integration, which compensates for the decline in neural efficiency associated with aging (Koen & Rugg, 2019; Park & Reuter-Lorenz, 2009). The stronger FC between the left IFJ and the left IPL as well as the right precuneus in older adults may indicate reduced specificity and efficiency, thereby leading to poorer task performance.

Limitations of the present study should be mentioned. First, participants in the current study were healthy adults aged 18–35 years and 60–80 years, and it is uncertain whether the current results can be generalized to other groups. Future studies are warranted to further investigate the validity of Agile Fisherman at the neural level in other populations, such as adolescents, middle-aged adults, or clinical populations with mild cognitive impairment or Alzheimer’s disease. Second, switching between two different tasks, such as the current Agile Fisherman, has been a common paradigm for assessing cognitive flexibility in neuropsychological assessments (Diamond, 2013). However, it may not capture the full picture of cognitive flexibility, but only one specific aspect, the ability to flexibly switch between two specific task requirements. The underlying mechanism of cognitive flexibility cannot be fully elucidated by traditional experimental and modeling studies that are constrained to focus on one or very limited number tasks at a time. Future studies are expected to further explored the brain mechanisms underlying cognitive flexibility through more sophisticated methods (Cabral et al., 2017; Capouskova et al., 2023; Yang et al., 2019a, b).

Conclusion

The current study finds that the key region responsible for cognitive flexibility, the left IFJ, is significantly activated during gamified cognitive flexibility tasks and correlated with behavioral performance in both young and older adults. Moreover, the results showed that the task was age sensitive, and the IFJ PSC, and representation precision in older adults were higher than those in young adults under repeat and switch conditions. The current findings demonstrate the effectiveness of Agile Fisherman at the neural level, highlight the necessity of brain imaging validation, and provide a promising approach for future validation of newly developed cognitive assessment tools at the neural level.

Fig. 1
figure 1

Flow chart of the functional magnetic resonance imaging (fMRI) paradigm of Agile Fisherman. Color task: if the middle net has a handle, participants need to catch according to the color of the net; Shape task: if the middle net does not have a handle, participants need to catch according to the shape of the net. At the time of scanning, participants use a custom-designed response pad that is compatible with fMRI to complete the task. There are 8 buttons in the response pad, corresponding to the bubbles in the corresponding direction

Full size image
Fig. 2
figure 2

Functional activation in older and young adults during the game. (a) Left inferior frontal junction (IFJ), inferior parietal lobule, and superior parietal lobule were significantly activated in older adults, and the left IFJ activation intensity was significantly correlated with the game performance. (b) The left IFJ, left middle frontal gyrus, and inferior frontal gyrus were significantly activated in young adults, and the left IFJ activation intensity was significantly correlated with the game performance

Full size image
Fig. 3
figure 3

Functional connectivity (FC) of the left inferior frontal junction (IFJ) in older and young adults during the task. (a) The left inferior parietal lobule (IPL) and the right precuneus had significant FC with the left IFJ in older adults. (b) The left middle frontal gyrus (MFG) and the right angular had significant FC with the left IFJ in young adults. (c) FC of the left IFJ with the both left IPL and the right precuneus was negatively correlated with the game performance in older adults. Note: L, left; R, right; VOI, volume of interest

Full size image