We then calculated light propagation in the subject’s head at wavelengths of 780, 805, and 830 nm for this optode arrangement. For constructing an optical model including the optodes’ precise positions, T1 weighted MR images of the subject both before and after implantation of the optode sockets were obtained. After, images were obtained by filling the socket wells with an MR marker (Cu-sulfate solution). On the basis of this optical model, SSP and the optical partial path length at each wavelength at each channel were calculated in the same way as described above.
Figure 3a shows the image of subject’s brain with optodes (lower center) and SSPs at the wavelength of 805 nm at 27 channels. In general, the SSPs of adjacent channels overlapped indicating that the optode arrangement offers nearly complete coverage of the underlying gray matter layer. The maximum value of SSPs from all of the channels was calculated (Fig. 3b), allowing for demarcation of areas of higher and lower sensitivities. For appropriately comparing hemodynamic responses among the channels, the fNIRS signals were calibrated using the estimated optical partial path length at each wavelength at each channel.
One hundred thirty milliseconds were required to complete illumination at all optodes while recording two measurements for every channel. We used the average of the two measured absorbance data as one sample in a temporal resolution of 130 ms. These absorbance data were calibrated using the optical partial path length estimated through the simulation described above. Baseline drifts in the calibrated absorbance data were removed by a third polynomial fitting. High-frequency noise was filtered using a fourth Butterworth-type filter at 0.7 Hz.
The concentration changes in oxygenated and deoxygenated Hb, ΔHbO and ΔHbR, respectively, were transformed from the calibrated absorbance data at each measurement wavelength on the basis of the modified Beer–Lambert law. The molar absorption coefficients of each Hb species in the translation matrix were taken from a literature34.
Event-related experiment of food retrieval task
The monkey was trained a small-object retrieval task closely resembling that used in our previous studies12,35. In this task, the monkey sat in the primate chair and retrieved a small spherical food pellet (5 mm in diameter) from the Klüver board, which contained cylindrical wells of three different sizes (10, 11, and 20 mm in diameter, 5 mm in depth). The Klüver board was located at the monkey’s waist height and at a sagittal distance of about 300 mm from the monkey’s shoulder. The size of the well was fixed within a daily session, and the monkey retrieved food pellets approximately every 20 s using each hand alternately. One hundred fifty trials (75 for each hand) were performed in each daily session. To observe behavioral changes during the learning of food retrieval, two separate series of food retrieval sessions were executed. In both the session series, a food well of 11 mm in diameter was used for the initial session. Food wells of 20 and 10 mm were separately used for three additional sessions in each session series. Each session series was completed within a week. More than several weeks were left between the two session series.
Two additional sessions were also performed: (1) food retrieval only by left hand in 75 trials in which the subject’s right forelimb was constrained, (2) direct feeding to the mouth by experimenter in 75 trials with both forelimbs constrained. Cross-validation of potential contaminants in the fNIRS signals was conducted more than several weeks before of the start of the testing sessions and was performed using left and right hands alternatively.
The onset and end time points of food retrieval movement by either the left or right hand were detected by a Digital Laser Sensor (LV-11SB with sensor head LV-S72, Keyence, Osaka, Japan) installed in slits within the acrylic resin board located between the primate chair and the Klüver board (Fig. 2b). These retrieval event data were obtained concurrently with the fNIRS signals and were used for event-related signal analyses.
As described in the previous section, the onset and end time points of food retrieval movement by the left or right hand were monitored with the Digital Laser Sensor. The signal from the Digital Laser Sensor was received and recorded with the OMM-3000 concurrently with the fNIRS absorbance data. On the basis of this retrieval event data, the times of onset and duration of each retrieval event in each experimental trial were measured and used for the event-related signal analyses described below.