The Neural Markers of Perceptual Uncertainty/Curiosity—A Functional Near-Infrared Spectroscopy Pilot Study

The study investigates how cortical activation reflects the induction and relief of perceptual curiosity using a blurred picture paradigm adapted from fMRI to fNIRS. Curiosity is framed as a psychological state arising from recognition of an information gap and the expectation that closing this gap is valuable and possible. It serves exploratory, motivational, protective, and developmental functions, influencing learning, attention, decision-making, and well-being. Prior neuroimaging work implicates a network including OFC, ACC, IPL, dopaminergic midbrain, hippocampus, and striatum. The research question centers on whether fNIRS can detect cortical activity changes, particularly in OFC and inferior parietal lobule (IPL), during reduction of perceptual uncertainty (matching clear image after a blurred one) versus when uncertainty is not resolved (non-matching clear image). The pilot aims to validate methodological transfer from fMRI to fNIRS and expects OFC-related activation during uncertainty reduction.

Perceptual uncertainty, a component of perceptual curiosity, is elicited by ambiguous stimuli and reduced via exploration, drawing attention automatically. Blurred images have long been used to manipulate and study object recognition and emotional processing, with blurring shown to diminish affective modulation in physiological measures while leaving the startle reflex intact. Neuroscientific studies link perceptual uncertainty and curiosity to learning and decision-making circuits, including cortico-striatal-thalamic pathways and models of decision uncertainty. Prior fMRI work (Jepma et al., 2012) using blurred pictures reported increased activity in ACC and anterior insula during uncertainty induction and OFC and LOC during relief. Additional studies observed curiosity-related engagement of ACC, NAcc, parietal cortex, rlPFC (epistemic curiosity), vmPFC (surprise and valuation), and occipitotemporal cortex for uncertainty processing. The PACE framework (Gruber & Ranganath) integrates prediction errors, evaluation, curiosity, and exploration, involving hippocampus, ACC, lateral PFC, amygdala, and dopaminergic systems. OFC has been implicated in reward evaluation and relief conditions following curiosity induction. Some prior work demonstrated feasibility of assessing OFC activity with fNIRS, though spatial resolution and sensitivity constraints necessitate combined measurement of neighboring prefrontal structures.

Design: Pilot study adapting the blurred pictures fMRI paradigm to fNIRS. Four trial types: (1) blurred → clear matching; (2) blurred → clear non-matching; (3) clear → blurred matching; (4) clear → clear matching. Experimental comparisons focused on conditions (1) vs (2) after the second image; (3) and (4) served as distractors. Stimuli and timing: Images from Rossion & Pourtois standardized colored object set. Each participant viewed unique, non-repeating pairs (80 distinct images and blurred counterparts), randomized across trials and conditions. Procedure: welcome/instructions, demo pair, fixation baseline (30 s), then 10 randomized trials per participant. Each trial: Image 1 (10 s) → fixation (0.5 s) → Image 2 (10 s) → intertrial interval (random 8–13 s). Extended presentation windows were chosen to capture a fuller hemodynamic response suitable for fNIRS while balancing session length and participant comfort. Participants: N=15 (14 females, 1 male), age 21–30, at least high school education, screened for visual/neurocognitive impairments. Exclusion criteria: prior brain damage or implants; neurological or psychiatric disorders; uncorrected vision; fine motor difficulties; significant scalp damage; regular medication use. Informed consent obtained; study approved by ethics committee (protocol 18/2023). fNIRS acquisition: NIRX Scout system, two wavelengths (760, 850 nm), measuring changes in oxygenated/deoxygenated hemoglobin. Probe: 30 emitters, 22 detectors arranged in 10–20 EEG positions covering frontal, parietal, temporal cortices; 90 channels; source-detector distance ~30 mm. Optode localization and channel sensitivity estimated via fOLD toolbox (Brodmann atlas referencing). Regions-of-interest included channels with specificity to BA11 (OFC) and BA10 (frontopolar/rostrolateral/anterior PFC): channels 12, 15, 17, 54, with BA11 specificity ~20–45% and predominant BA10 coverage. Procedure and environment: Participants seated ~50 cm from LCD monitor; fNIRS cap calibration and optode positioning; shaded room to reduce light interference; continuous presence of researcher during data collection. Signal processing: Conducted in Homer3. Steps: raw signal → optical density (Beer–Lambert); bandpass filter (HPF 0.010 Hz; LPF 0.200 Hz); motion artifact detection (tMotion=0.5; tMask=1; STDEVthresh=50; AMPthresh=5.0) and spline interpolation correction (p=0.99); conversion to concentration changes (hmrR_OD2Conc). HRF estimation via GLM: trange −2 to 10 s; glmSolveMethod=1; idxBasis=2; paramsBasis=1.0, 1.0; rhoSD_ssThresh=15.0; flagNuisanceRMethod=0; driftOrder=3; c_vector=0. Eight regressors modeled stimulus onsets (two images × four conditions) plus one for intertrial intervals. Quality assessment via QT-NIRS with SCI threshold 0.8, PSP threshold 0.1, minimum signal quality 75%. One participant excluded due to low signal quality across most channels; parietal channels were removed due to high noise. Statistical analysis: For Oxy-Hb β coefficients from the GLM, paired-sample t-tests compared conditions (1) matching vs (2) non-matching after unveiling the second image, per channel (12, 54, 15, 17). Analyses performed in IBM SPSS 29.

• After exclusions, N=14 participants analyzed. Focused channels: 12, 54, 15, 17 with OFC/BA10 sensitivity.
• Paired t-tests (matching vs non-matching after second image) on Oxy-Hb β values: • Channel 12 (AF3–Fp1; left frontal, ~20% BA11/OFC specificity): mean β matching = −0.000032959; mean β non-matching = −0.000118106; SD = 0.000122152; t(13) = 2.284; p = 0.04 (two-tailed). Indicates higher activity when uncertainty was reduced (matching condition). • Channel 54: t(13) = 0.117; p = 0.91 (ns). • Channel 15: t(13) = 1.320; p = 0.21 (ns). • Channel 17: t(13) = −1.050; p = 0.31 (ns).
• Parietal channels were excluded due to noise; thus IPL hypotheses could not be tested.
• Overall, evidence of a differential hemodynamic response consistent with OFC-related engagement during uncertainty reduction was observed in one left prefrontal channel (12).

Findings partially support the hypothesis that OFC and adjacent prefrontal regions show increased activation when perceptual uncertainty is relieved (matching blurred-to-clear images). The significant effect in channel 12 aligns with prior fMRI evidence implicating OFC in reward evaluation and relief from uncertainty, and with broader curiosity circuitry (OFC, ACC, NAcc, parietal cortex). OFC’s role in emotion regulation, reward processing, and its connectivity with striatal circuits suggests involvement in monitoring reward-related signals when curiosity is satisfied. Given channel 12’s predominant coverage of BA10 along with partial BA11 specificity, contributions from dorsolateral, dorsomedial, and ventromedial prefrontal regions may also underlie the observed effect, implicating affective memory regulation and evaluative processes. Methodological constraints of fNIRS (limited spatial resolution, depth sensitivity, and difficulty isolating OFC from adjacent PFC) temper localization claims. Habituation and lack of interactivity may have attenuated curiosity induction over an extended session, potentially averaging out activation differences. Alternative interpretations include compensatory involvement of other cortical regions (e.g., DLPFC), but subcortical structures like the amygdala and hippocampus, relevant to curiosity/reward/memory, remain inaccessible to fNIRS. Demographic variables (gender imbalance, narrow age range, educational background) may contribute to individual variability, though literature generally reports minimal gender effects on curiosity and mixed findings regarding age and education. The pilot’s primary contribution is demonstrating feasibility of adapting an fMRI curiosity paradigm to fNIRS with preliminary evidence of OFC-related activation during uncertainty reduction.

This pilot study adapted an fMRI-based blurred picture paradigm to fNIRS and provided preliminary evidence of greater OFC-related activity during the reduction of perceptual uncertainty (matching condition) compared to non-matching. While not all hypotheses were confirmed and effects were limited to a single channel, results are consistent with proposed curiosity pathways. For future curiosity research, the authors recommend optimizing experimental design to mitigate habituation (e.g., fewer trials, interactive elements) and employing larger samples with stricter statistical thresholds to enhance robustness. The work underscores both the promise and spatial limitations of fNIRS for studying OFC involvement in curiosity.

• Small sample size (N=14 analyzed) increases sensitivity to individual variability and limits statistical power.
• Most analyzable channels did not exhibit significant differences; effect observed only in channel 12.
• Inferior parietal lobule (IPL) ROI was excluded due to unsatisfactory signal quality, limiting tests of parietal involvement.
• Potential measurement issues: suboptimal cap fit for some participants; possible insufficient sampling frequency leading to aliasing; general fNIRS spatial/depth limitations hinder precise OFC localization.
• Habituation and lack of interactive engagement may have reduced manipulation efficacy over the long session.
• Exploratory pilot without correction for multiple comparisons; the observed p=0.04 is near the conventional threshold and should be interpreted cautiously. Future work should use more conservative thresholds (e.g., p<0.01) and larger samples.