This study was conducted from April 15 to July 12, 2016. We recruited a volunteer group of the healthy, right-handed older adults to participate in the study through the Social Welfare Council in Tatsuno City, Hyogo Prefecture, Japan. Eighteen members of the Asatomi senior citizen’s club in Tatsuno City offered their cooperation for the study. All of them lived independently at home and had not been received a diagnosis of having mild cognitive impairment or dementia. The participants were 18 right-handed healthy older adults (M = 71.7 years ± SD 5.0, range: 63 - 84 years) of which two were males (M = 74.5 years ± SD 0.71, range: 74 - 75 years) and 16 were females (M = 71.4 years ± SD 5.3, range: 63 - 84 years). Their right-handedness was confirmed in the oral interview and through observation of their hand motion during the fNIRS measurements.

Watering seemed to be appropriate as a gardening task for the participants easy to do and continue at home. We requested the participants to continue the watering activity at home for three months and measured the blood flow in the FP during the watering task using fNIRS before and after the intervention period. The way of watering at home was the same as the watering task at measurement (Figure 1).

Statistical analyses were performed using Excel 2010 (Microsoft, USA) with the add-in software SSRI, Version 1.02, 2012 (Social Survey Research Information Co., Ltd.).

The participants’ task performance showed the similar tendency both in the pre- and the post-intervention measurements. The task performance in the second and the third trials improved compared with the first trial (Figure 4). In addition, the task performance in the first trial in the post-intervention measurement was higher than that in the pre-intervention measurement (Figure 5). From the above, it can be said that the watering task performance of the older adults improved after three-month intervention.

Prior to executing the watering task, the participants were requested to memorize the prescribed way of watering according to the instruction. The watering task consisted of three processes: 1) to water the soil in a cell in a cell tray, 2) to raise the plastic bottle upright and 3) to move to the next cell to water.

During the task performance, attention to visual stimulation such as a cell, a plastic bottle, pouring water, and somatosensory stimulation such as the weight of the plastic bottle containing water and the tilt of the plastic bottle are necessary. Burgess, Simons, Dumontheil, and Gilbert described a function of medial BA10 [Brodmann area 10] as “biasing attention toward current sensory input” [19] . Oxy-Hb values at CH 7 and CH 10 are considered to reflect the medial FP activation due to biasing attention toward current sensory input, namely SO (stimulus-oriented) attention/thought.

Memorizing a set of instructions about how to water, recalling individual action at each process and thinking of the next action to be taken during task performance are presumed to correspond to SI attending [15] or SI thought [16] . The lateral FP is considered to be “activated bilaterally during periods of extended stimulus-independent cognition” [20] . Thus, Oxy-Hb values at CHs 4, 5, and 6 in the right hemisphere, and CHs 11, 12, and 13 in the left hemisphere seem to reflect the lateral FP activation during SI thought.

Different functions have been reported for the lateral FP in each hemisphere. Gilbert, Frith, and Burgess reported transient activity in the right lateral rostral PFC by a switch between SO and SI thought [16] . In our study, attending to visual and somatosensory stimulation corresponds to SO attention/thought. Switching attention between external stimulation (such as visual and somatosensory stimulation) and internal thought (such as thought of the next action) is considered as switching attention between SO attention and SI thought. Oxy-Hb values at CHs 4, 5, and 6 also reflect the right lateral FP activation during switching attention between SO attention and SI thought.

Pollmann indicated selectively increased dimension-change-related activation in the left FP for stimulus-driven visual dimension weighting [21] . Changes in the participant’s visual weighting during watering are thought to be visual dimension changes: As watering proceeds, the participant’s focus moves, within a certain sight, to a frame of the cell, the nozzle of the plastic bottle, the running water from the nozzle, and the rising level of water in the cell sequentially. Oxy-Hb values at CHs 11, 12, and 13 also reflect the left lateral FP activation during visual dimension change.

Focusing on CH 7 and CH 10 located in the medial FP, the tendency of activation was similar both in the pre- and the post-intervention fNIRS measurements in that the maximum and the mean Oxy-Hb values in the first trial were the highest of the three trials. However, as we mentioned in Section 3.2.1 and Section 3.2.2, comparison of the maximum and the mean Oxy-Hb values among three trials in the medial FP did not show a statistical decrease in the second and the third trials compared with the first trial in the pre-intervention measurement. This result was the same as that in our previous study where the healthy older adults repeated the watering task five times [11] . On the contrary, in the post-intervention fNIRS measurements, the maximum and the mean Oxy-Hb values showed a statistical decrease in the second trial at both channels compared with the first trial.

The reason why there was no significant change among three trials in the pre-intervention is presumed that the watering task required coincident attention to two types of external input (i.e. visual information, and somatosensory information) and the medial FP activation was necessary even after the second trial to learn a suitable motion for performing the task. However, in the post-intervention measurement after three months (i.e. after putting watering into practice over one thousand times), the participants were well accustomed to the external input during watering. As a result, the medial FP did not need the similar activation any longer to do the same action.

As we mentioned in Sections 3.2.1 and 3.2.2, the activation of the lateral FP showed a similar declining tendency to that of the medial FP: The maximum and the mean Oxy-Hb values were the highest in the first trial and decreased in the second and third trials both in the pre- and the post-intervention measurements. The reason why the number of channels showing a significant decrease in activation increased in the second trial in the post-intervention measurement is presumed that the participants got accustomed enough to watering. As a result, they needed less SI cognition for performing the task compared with the pre-intervention measurement.

In spite of a gradual performance improvement during three trials in the pre- and post-measurements, the FP activation in the second or the third trial showed a declining tendency in both measurements. This indicates that the task demands of cognitive function in watering were not so high. Our results support the idea proposed by Strange, Henson, Friston, and Dolan: “we suggest that the FPPC (fronto-polar prefrontal cortices) is engaged during intentional or explicit rule induction but once a rule is learnt, more posterior prefrontal areas mediate rule application” [22] .

In addition, we would like to discuss the FP activation related to characteristics of watering: Even if the participants are well accustomed to the watering task after a three-month actual practice (estimated over 1000 times) at home, activation in both the medial and the lateral FP were observed at first during executing the task in the prescribed way another day. This result indicates a possibility that watering in a prescribed, same way may activate the FP whenever the participants start watering.

Concerning the reasons of the suppression of activation decline at CH 4 and CH 5 in the third trials within the post-measurement, first, we discuss a possibility of age-related compensation.

Cabeza and Dennis proposed a model of age-related compensation in PFC activation [23] . This model suggested that mismatch between available cognitive processing resources and task demands triggers the recruitment of reserve neural resources to increase the neural supply. They regarded an increase in neural supply that may enhance cognitive processing as “successful compensation” and the increase in neural supply that may not enhance cognitive processing as “unsuccessful compensation”. Is there a possibility that an age-related decline in the lateral FP function over the course of a three-month study period induced a greater activation in the right lateral FP than the pre-intervention measurement to perform the same watering task?

In our study, both in the pre- and the post-measurements, the task performance in the second and the third trials improved compared with the first trial, while the Oxy-Hb values during the watering task were the highest in the first trial and decreased in the second and the third trials. This result indicates that the task demand of watering was not high for the participants and the suppression of activation decline in the right lateral FP in the post-measurement was not caused by the mismatch between available cognitive processing resources and task demands. Therefore, it is hard to consider the suppression of activation decline in the right lateral FP as an age-related compensation.

Next, let us discuss a possibility of enhancement of cognitive function by repeating the watering activity.

In this study, the FP activation during the second and the third trials showed a declining tendency. The main reason seems to have been the participants’ getting used to the watering process. As a result, the participants needed less FP activation than the first trial. Moreover, in the post-measurement, the task performance improved.

The increase in the number of times of watering per trial means an increase in the number of switching between SO attention and SI thought within 15 seconds. The result simultaneously means that the ability to continue switching attention between SO attention and SI thoughts throughout a certain period of time required for the task improved. In other words, a three-month watering practice with a plastic bottle at home worked as a training for continual switching attention between external stimuli and internal thought. As a result of the actual watering activity, a favorable change attributable to brain plasticity that the activation in the right lateral FP was hard to decrease was showed confirmed in healthy older adults.

Reijnders, van Heugten, Boxtel indicated in their systematic review on cognitive intervention studies targeting healthy older adults that cognitive training (CT) can be effective in improving various aspects such as executive functioning and attention [24] . Mozolic et al. reported improvement of attention function in their randomized controlled trial (RCT) concerning the effects of CT targeting healthy older adults [25] . CT is defined as “an intervention providing structured practice on tasks relevant to aspects of cognitive functioning” [26] . Gates and Valenzuela also developed an operational definition of CT including components such as repeated practice and use of standardized tasks. They mentioned a trend that “traditional pen-and-pencil exercises are being increasingly replaced by computer-based programs” [27] . Meanwhile, Farzana, Ahuja, and Sreekanth included gardening as one of intellectually stimulating activities effective in cognitive stimulation therapy targeting people with mild to moderate dementia [28] .

Watering, one of gardening activities, is not CT intervention using traditional types of CT exercises or computer-based programs. However, our findings suggested that watering has a possibility to contribute to improving attention function of healthy older adults.

As we mentioned in Sections 3.2.1 and 3.2.2, some channels that showed decreased activation were different between the pre- and the post-intervention measurements. Jenkins, Brooks, Nixon, Frackowiak, and Passingham pointed out that the neural networks before and after proficiency in skills are different. They mentioned that “Comparing new learning with performance of the prelearned sequence, differences in activation were identified in other areas” [29] . In the course of the three-month continual watering, there might have been some kind of changes in the activated areas.

In this study, two possibilities were suggested: even the activity with a small number of steps such as watering, if the activity is continued for a certain period of time, can suppress the decrease in activation in the right lateral FP of older adults, and improve task performance (i.e. the number of times per unit time) with frequent switching between SI and SO attention.

Gardening involves many kinds of work such as seeding, weeding, deadheading, harvesting, and so on other than watering. All of them require switching between SO and SI attention. Then, what kind of brain activities in the FP take place during actual gardening? First, let us think about a case where the task-related routine is easily learned. Assuming that pots with well-grown seedlings are evenly arranged, when a person waters them in the same manner every day, the FP activation will occur every time at the beginning of watering. However, the FP activation is deemed to decrease in a relatively short time because the person will get accustomed to the work very soon. Next, let us consider harvesting cherry tomatoes. While a person is harvesting ones, he/she needs to adjust the posture of hands, feet and the whole body according to the situation of the plant in front of him/her. In order to harvest red ripe fruits at the appropriate harvest time, the operation performed for each plant varies every plant in order not to damage other tomatoes within the bunch and/or leaves around the target bunch. In this case, it will take more time to get used to the work than the former case. In addition, a slightly different motion is required for each plant according to the degree of growth of the plant and ripeness of the fruit. As a result, each time the person moves to the next plant to harvest, the FP activation will be repeated as is the case with working at a new task.

Some epidemiological studies reported that people who were engaged in gardening had a lower prevalence of Alzheimer’s disease [5] [6] [7] . These findings may indicate that continuous physical exercise by gardening contributed to the prevention of Alzheimer’s dementia, or that the repetitive motions accompanied by switching between SO attention and SI thought induced the continuous activation in the medial and lateral FP and contributed to the suppression of cognitive decline. Our findings showed a possibility that continuation of watering with only three processes for three months may contribute to the suppression of cognitive decline of healthy older adults. Moreover, continuation of gardening activities combined with different types may provide more effective stimulation to suppress cognitive decline.

In addition, in this study targeting older adults, a continuous gardening activity conduced to improvement in task performance accompanied by suppression of a decrease in the right FP activation. Low activation in the prefrontal area is observed not only in healthy older adults but patients with schizophrenia [30] . In the fields of occupational therapy, horticultural therapy, and green care, there are many cases where agricultural work and gardening programs are provided targeting patients with schizophrenia or people with intellectual disabilities. Further studies to investigate the relationship between the FP activation and task performance when agricultural work or gardening programs including repetitive motions are provided to such persons, will be meaningful.

In this study, we targeted older adults only. Evidence from younger adults and comparative discussion about the characteristics of the FP activation might be helpful to further figure out the training effect on FP activation by watering.

The fNIRS devise that we used in the present study measures the activation in the prefrontal area such as the whole FP, and portions of the DLPFC, the inferior frontal gyrus (including the orbitofrontal cortex), the pars triangularis Broca’s area, and the frontal eye field. As a result, we could not investigate the whole activation of prefrontal area except the FP. Verification using multi-channel fNIRS will also be required in the future research.