Stimulus-Related Neuroimaging in Task-Engaged Subjects Is Best Predicted by Concurrent Spiking

Abstract

The implicit goal of functional magnetic resonance imaging is to infer local neural activity. There is considerable debate, however, as to whether imaging correlates most linearly with local spiking or some local field potential (LFP) measurement. Through simultaneous neuroimaging (intrinsic-signal optical imaging) and electrode recordings from alert, task-engaged macaque monkeys, we showed previously that local electrophysiology correlates with only a specific stimulus-related imaging component. Here we show that this stimulusrelated component-obtained by subtracting a substantial task-related component-is particularly linear with local spiking over a comprehensive range of response strengths. Matches to concurrent LFP measurements are, to varying degrees, poorer. As a control, we also tried matching the full imaging signal to local electrophysiology without subtracting task-related components. These control matches were consistently worse; they were, however, slightly better for gamma LFP than spiking, potentially resolving discrepancies between our findings and earlier reports favoring LFP.

Laminar differences in responses to naturalistic texture in macaque V1 and V2

“…Previous BOLD experiments are clearly best matched to the multiunit spiking analyzed here in that V1 sensitivity is greater than 0 but a substantial difference in sensitivity remains between V1 and V2. Our results are thus consistent with recent studies that find a better match between multiunit activity and BOLD than between LFP and BOLD (Lima et al, 2014), and the use of naturalistic texture may thus provide a tool for further studying the relationship between different recording modalities.…”

Assessing sensory versus optogenetic network activation by combining (o)fMRI with optical Ca2+ recordings

“…Among other signal components, BOLD integrates both spiking as well as subthreshold potentials, and their relative contributions to the signal may vary. 21,49,50 Consequently, BOLD and Ca 2+ recordings provide complementary readouts of neural network activity.…”

Spatial Distribution of Attentional Modulation at Columnar Resolution in Macaque Area V4

“…Recently, Sirotin and co-workers have revealed that hemodynamic signals contain substantial task-related but stimulus-independent components that are not linked to neural activity, such as LFP and spiking activity (Sirotin and Das, 2009; Sirotin et al, 2012). Instead, stimulus-related components of CBV-based hemodynamic signals, which are obtained by removing task-related components with blank subtraction, are correlated linearly with spiking activities (the median determination coefficient R 2 = 0.83) more effectively than with LFP measurements (Cardoso et al, 2012; Lima et al, 2014). In this study, we also used blank subtraction to extract stimulus-related components in different attentional conditions.…”

“…We performed ISOI at 570 nm wavelength (an isosbestic, or equal absorption point of oxy- and deoxyhemoglobin), which emphasizes changes in cerebral blood volume (CBV; Frostig et al, 1990; Malonek et al, 1997). CBV-based hemodynamic signals are well correlated with neuronal activities in the cortex, including spiking activities and local-field-potential (LFP) measurements (Sheth et al, 2003; Nemoto et al, 2004; Niessing et al, 2005; Lima et al, 2014). We visualized maps of global and mapping signals of stimulus-evoked CBV-based hemodynamic responses under different attentional conditions and evaluated how these two types of signals were affected by attention.…”