Contrast response in visual cortex: Quantitative assessment with intrinsic optical signal imaging and neural firing

Abstract

While previous studies showed that intrinsic optical signals spatially correspond with electrophysiological responses in mammalian visual cortex, the quantitative correspondence of their response strengths is open to question. Measurement of both signals' strength as functions of visual stimulus contrast provides an opportunity for quantitative comparison. Towards that end, the spatial and temporal properties of the optical signal impose important constraints upon quantification of its strength. We used intrinsic optical signal imaging and single unit recording to measure responses to drifting gratings at contrasts ranging from 10–80% in cat area 18. We calculated the average difference images for pairs of oppositely moving, or orthogonally oriented, gratings at each contrast and evaluated three different methods for quantifying optical signal strength. After about 2.5 s, the spatial patterns of optical images and the time course of their strength were contrast-invariant. This “space–time-contrast separability” for optical response implies a spatial uniformity of the optical contrast response functions, provides an objective basis to guide temporal averaging of optical signals, and validates a scalar metric of optical signal strength. Optically measured contrast response functions increase monotonically and saturate at high contrasts, qualitatively resembling those from single units. However, quantitative comparison reveals a nonlinear relationship with neural firing, such that the optical response reaches half of its maximum when the neural response has reached only around 20% of its maximum. This relationship suggests that intrinsic optical signals are relatively more sensitive to weak signals than neural firing.

Introduction

Intrinsic optical signal imaging measures local activation-related changes in light reflected from neural tissue (for review, see Blasdel, 1997, Grinvald et al., 1999). While this functional imaging was validated by electrophysiology (Shmuel and Grinvald, 1996), supporting an implicitly accepted view that the former represents neural firing rates, the indirect nature of the optical signal raises questions of its quantitative relation to neural activity and the coupling mechanism. To resolve these issues requires a better understanding of the spatial and temporal properties of the optical signals and of how to appropriately quantify their magnitude. Similar issues have been of recent interest for blood-oxygenation-level-dependent (BOLD) fMRI where a quantitative understanding of the relationship is also crucial for interpreting imaging results functionally (e.g., Logothetis et al., 2001, Rees et al., 2000). Because the intrinsic optical signal is hemodynamically driven and reflects oxygenation (Malonek and Grinvald, 1996), its relationship to neural firing may also be relevant to the interpretation of fMRI signals.

Previous efforts to compare neuroimaging with electrophysiology have concentrated on their spatial correspondence, for optical imaging, or response strengths, for fMRI. Shmuel and Grinvald (1996) and Masino (2003) demonstrated that optical signals are spatially correlated with electrical activity. Thompson et al. (2003) found correlated changes of local oxygenation and neural firing. A linear proportionality between monkey neuron firing rates and human fMRI strength was reported by Heeger et al. (2000) and Rees et al. (2000) for varying grating contrast or random-dot motion coherence, respectively. However, Logothetis et al. (2001), recording simultaneously the response to a series of contrast stimuli from monkey V1, showed that fMRI is quite nonlinearly related to multi-unit firing. This nonlinearity might arise from a disproportionate contribution of metabolic demands and/or subthreshold activity (Das and Gilbert, 1995, Toth et al., 1996) to imaging signals.

The visual cortex of the cat (Payne and Peters, 2002) affords a good opportunity to quantitatively compare optical imaging and neural firing—it is rich in neurons selective for orientation and direction of motion, columnar organization of both properties has been revealed by optical imaging (Shmuel and Grinvald, 1996, Weliky et al., 1996) as well as electrophysiology (Swindale et al., 1987), and the contrast response functions (CRFs) of single units have been extensively studied (e.g., Albrecht et al., 2002, Bonds, 1991). Therefore, in this preparation, it is feasible to compare differential responses to opposite directions of motion, or orthogonal orientations, in a parallel manner for optical signals and for neurophysiology, as functions of contrast.

We have recorded and systematically quantified CRFs for orientation and direction selectivity by both optical imaging and single unit recording in cat A18. The spatial patterns of optical images and the time course of optical signal strength are contrast-invariant, implying a “space–time-contrast separability” of the optical signals. This property implies a spatial uniformity of the optical contrast response function, provides an objective basis to guide temporal averaging of the optical signals, and validates a scalar metric of optical response strength. Both optical imaging and electrophysiological CRFs are monotonically increasing functions, but the former saturates at lower contrasts. These results suggest that hemodynamically driven imaging signals are relatively more sensitive to weak signals when measured using a differential protocol and represent a more complex signature of neural activity than simply neural firing. Some aspects of this work have been described in abstract form (Zhan et al., 2002).