Is Light Part of the Future of Precision Psychiatry?

Psychiatry faces a core challenge: We treat disorders of the mind with limited windows into real-time brain function. Unlike cardiology’s continuous monitoring (ECG, ultrasound, nuclear imaging), mental health care relies mainly on subjective reports, behavioral observations, standardized measures, and occasional scans that provide snapshots rather than movies. Real-time imaging, such as white-matter tractography, is cumbersome and expensive.

Functional near-infrared spectroscopy (fNIRS) could help bridge this gap—not as a total solution, but as a tool to link brain activity patterns with mental function for treatment planning, monitoring, and fine-tuning. Key advantages include portability, movement tolerance, and safety for repeated measures. Unlike fMRI, fNIRS can be worn during naturalistic behavior, improving ecological validity, and may be less sensitive to artifacts in some regards than EEG.

Compared with EEG, which captures fast electrical signals but with limited spatial resolution, fNIRS provides finer spatial specificity for cortical hemodynamics at slower timescales. Used together, they offer accessible tools that can support precision interventional psychiatry.

How light reveals brain activity

fNIRS exploits the fact that oxygenated and deoxygenated hemoglobin absorb near-infrared light differently. Light sources and detectors on the scalp emit and capture wavelengths roughly between 650–1,200 nm, penetrating into cortical tissue about 2–3 cm (Chen et al., 2020). When neurons activate, oxygen use rises, followed by a compensatory increase in blood flow that typically overshoots demand.

This "neurovascular coupling" yields increased oxygenated and decreased deoxygenated hemoglobin at active sites, allowing inference about which cortical areas are engaged during tasks or mental states. fNIRS devices produce a lot of data, as well. What we do with that, leveraging increasingly sophisticated machine learning and AI approaches, is very promising. As it stands, we can see a real-time snapshot of cortical activity.

Source: High-Density fNIRS, NIRSIT, Soterix

Crucially, more blood flow does not mean better or worse mental health; meaning lies in patterns, networks, and context. Individuals with major depression, schizophrenia, and bipolar disorder often show prefrontal/frontotemporal hypoactivation during verbal fluency and executive tasks (Ehlis et al., 2014; Yeung & Lin, 2021). Clinical acumen remains foundational so that tools like fNIRS can best serve patients' needs.

Enhancing therapeutic understanding

fNIRS could be useful not only for live viewing during treatment, but also for between-session analysis and planning. It could test whether interventions engage hypothesized neural mechanisms. For example, does cognitive restructuring strengthen prefrontal regulatory circuits? Does exposure therapy reduce hyperreactivity or merely produce surface compliance?

Neurofeedback work using fNIRS and related consumer-grade tools shows promise (Flanagan & Saikia, 2023). Teaching patients to modulate their own brain activity blends contemplative practices with modern neuroscience, offering additional self-regulation tools while preserving the centrality of the therapeutic relationship.

In trauma-related disorders, measuring prefrontal-limbic connectivity patterns over time may indicate strengthening of regulation circuits. Objective feedback could guide pacing of trauma work and the integration of stabilization techniques, complementing clinical judgment.

Precision in psychiatric medication

Neuroscience Essential Reads

Ultrasound and the Brain

The Downside of Silence

Medications don't work as well as we'd like; transcranial magnetic stimulation (TMS), for example, vastly outperforms polypharmacy in patients receiving psychotherapy. fNIRS may inform medication selection and monitoring by detecting early treatment response patterns. Changes in prefrontal activation during cognitive tasks can sometimes predict antidepressant response before symptoms improve (Wei et al., 2021; Băcilă et al., 2025). Earlier prediction could shorten ineffective trials. In clinical depression, for example, researchers (Ho et al., 2025) found that fNIRS with machine learning was very effective in measuring dorsolateral prefrontal cortex (dlPFC) cognitive task-evoked changes to predict six-month clinical outcomes.

Diagnostic differentiation is another avenue. Although bipolar disorder and major depression can present similarly, fNIRS studies report distinct hemodynamic patterns during tasks (Yeung & Lin, 2021). Machine learning approaches applied to fNIRS have achieved promising classification performance, including multi-class classification of neuropsychiatric conditions (Erdoğan & Yükselen, 2022). Still, overlaps between conditions, heterogeneity, and task differences limit clinical specificity.

Optimizing neuromodulation therapies

Treatments like TMS help many with treatment-resistant depression, especially more recent "accelerated" protocols (aTMS), but response is variable and hard to predict. fNIRS could improve targeting by mapping individual activity patterns to account for anatomical and functional variation, then monitoring response. During treatment, fNIRS can test target engagement: Is the intended network modulated? Early work suggests fNIRS-derived patterns may predict TMS response, potentially sparing non-responders lengthy, ineffective courses (Băcilă et al., 2025). Similar principles could guide transcranial direct current stimulation and focused ultrasound, where precise targeting matters.

A network perspective is essential. Many mental health conditions reflect disrupted connectivity rather than simple over-/under-activation of isolated regions. fNIRS can measure functional connectivity among cortical regions, illuminating network-level disturbances beyond single-region activation.

Supporting behavioral interventions

Lifestyle factors shape mental health, but adherence is hard. fNIRS could provide objective evidence of how exercise, meditation, sleep, and social connection affect cortical function, potentially motivating adherence through visible progress. Neuromodulation can enhance psychotherapy, "bending the developmental curve."

Exercise is a case in point. Aerobic activity can rival antidepressants for some, yet we lack predictors and dose guidance. fNIRS could show how intensity and duration shift prefrontal-limbic dynamics, enabling personalized prescriptions and creating a feedback loop where awareness of neural changes reinforces healthy behavior.

fNIRS is especially promising in pediatrics, where traditional imaging is challenging and early intervention matters (Obrig, 2014). Adolescents may engage more with technology-enabled, objective feedback about how coping strategies alter brain responses.

Integration, ethics, and clinical reality

As fNIRS matures, integration with digital therapeutics and wearables could allow more frequent sampling of neural states outside clinics. This raises questions about neural data privacy, interpretation, and consent. While cheaper than MRI, fNIRS still requires investment; implementation will determine whether it reduces disparities or widens them.

Limitations are clear. fNIRS does not capture subcortical structures (e.g., amygdala, hippocampus, striatum) and is limited to superficial cortex. Methodological variability and lack of standardized protocols hinder clinical adoption (Băcilă et al., 2025; Obrig, 2014). In addition, using dynamic causal modeling is important to find causal, and not merely correlational, relationships.

Within these constraints, fNIRS offers practical, repeated, naturalistic measures of cortical function. Aligned with the NIMH Research Domain Criteria framework, which emphasizes dimensional biology over categorical diagnoses, fNIRS could help identify transdiagnostic neural markers to provide updated ways to diagnose and treat psychiatric and neurological conditions, with an empirical basis, based on brain activity, in addition to psychosocial factors.

Bridging subjective and objective realms

fNIRS alone will not solve psychiatry’s complexities. Subjective experience cannot be reduced to hemodynamics, and the therapeutic relationship remains central. But as a bridge between subjective reports and objective measures, fNIRS can help validate experience and track treatment effects.

Advances in machine learning and signal processing are improving the extraction of clinically meaningful information. Combined with clinical interviews, behavioral observation, and digital phenotyping, fNIRS could support genuinely personalized psychiatric care.

The integration of fNIRS into mental health care represents evolution, not revolution—incrementally improving our capacity to see, understand, and help. In narrowing the brain-mind gap, we do not diminish the mystery of consciousness; we gain better tools to address its disruptions. For clinicians, researchers, and, most importantly, patients, fNIRS offers not definitive answers but better questions.