1. The Blood-Oxygenation-Level Dependent (BOLD) effect does not measure neuronal activity directly.

2. The BOLD signal reflects changes in blood oxygenation resulting from the relative balance between cerebral blood flow (oxygen supply) and oxygen metabolism (oxygen consumption).

3. Cerebral blood flow and oxygen metabolism responses to brain activation can be accurately obtained through precisely calibrated fMRI studies.

Researchers and clinicians interested in getting a fundamental understanding of the physiological sources underlying the most commonly used fMRI BOLD signal for detecting brain activity.Develop an awareness of the importance of robust experimental conditions to move from BOLD fMRI being a simple qualitative localizer (‘blobology’) to being a reliable quantifier of cerebral activation.The BOLD fMRI signal detects a change in magnetic properties of haemoglobin in different oxygenation states. BOLD hence does not measure neuronal activity directly but reflects alterations in blood oxygenation resulting from the relative balance between oxygen (O₂) delivery and consumption (Fig1). Accordingly, one cannot rely only on this complex signal; one must also quantify cerebral blood flow (CBF) and cerebral metabolic rate of oxygen consumption (CMRO₂).¹ While the former can be measured using arterial spin labeling (ASL) to estimate the latter through an fMRI calibrated model involving changes in inspired levels of carbon dioxide (CO₂), conventional gas manipulation leads to estimation uncertainties. Recent advances in gas control produce more stable signals and reduce measurement variability to enlighten the intricate relationship between the haemodynamic and metabolic changes underlying the BOLD phenomenon.Careful and precise calibrated fMRI studies were performed with novel gas delivery methodologies in healthy adults at 3T during respiratory and neuronal tasks. We compared a robust computer-controlled gas system to 1) the traditional manual technique through randomized graded challenges of elevated CO₂ levels delivered via facemask, and 2) alternate calibration gas type of elevated O₂ challenges. Following each calibration, subjects performed simple visual and sensorimotor tasks, consisting of looking at an alternating radial checkerboard while performing voluntary bilateral finger-to-thumb apposition. The metabolic fMRI responses to neuronal activation were estimated based on acquired BOLD and CBF data under individual- and group-calibration in each activated brain region.The precise automated gas control yielded signals of increased linearity and uniformity across the brain, with reduced intra-/inter-subject variations, compared to the manual traditional technique (Fig2)² and under high O₂ rather than CO₂ calibration (Fig3)³. While most fMRI studies to date have been limited to group-calibration due to excessively large errors, reduced variability allowed accurate calibration of localized neuronal activation in individual subjects (Fig4) and, therefore, proper quantification of their distinct blood flow and oxygen metabolism responses in each brain region (i.e., visual and sensorimotor cortices, Fig5).⁴Novel gas delivery methodology provided robust quantification of the haemodynamic and metabolic responses underlying brain activation. Accurate evaluation of relative changes in CBF and CMRO₂ is vital in assessing the validity of BOLD as an indicator of neuronal activation under various healthy and physiological conditions. The demonstrated linear flow-metabolic relationship (Fig5) ensures a monotonic relation of the BOLD signal with these surrogates, greatly simplifying its applicability in the human brain.Despite the continuing debates surrounding which features of brain stimulation solicit higher O₂ metabolism and which mediate the flow response, one certainty remains in that the resulting BOLD phenomenon is critically sensitive to the exact coupling of these changes. Variations in the BOLD fMRI signal due to altered baseline states (e.g., neurovascular diseases, pharmacological manipulation, caffeine consumption and even attention) also play a crucial role in the interpretability of fMRI studies, hence necessitating precise calibration. Interpretation of BOLD results should be based on accurate measurements of underlying haemodynamic and metabolic responses, not only in basic neuroscience research but also in its countless clinical applications, where flow-metabolic coupling and baseline states may likely be impaired.