Event-related functional MRI of awake behaving pigeons at 7T

Abstract

Animal-fMRI is a powerful method to understand neural mechanisms of cognition, but it remains a major challenge to scan actively participating small animals under low-stress conditions. Here, we present an event-related functional MRI platform in awake pigeons using single-shot RARE fMRI to investigate the neural fundaments for visually-guided decision making. We established a head-fixated Go/NoGo paradigm, which the animals quickly learned under low-stress conditions. The animals were motivated by water reward and behavior was assessed by logging mandibulations during the fMRI experiment with close to zero motion artifacts over hundreds of repeats. To achieve optimal results, we characterized the species-specific hemodynamic response function. As a proof-of-principle, we run a color discrimination task and discovered differential neural networks for Go-, NoGo-, and response execution-phases. Our findings open the door to visualize the neural fundaments of perceptual and cognitive functions in birds-a vertebrate class of which some clades are cognitively on par with primates.

Fig. 1. Platform for awake pigeon fMRI.

a Head-restrained pigeons were positioned in the scanner bore center and stimuli were presented using fiber optics connected to a light stimulator outside the scanner room. Animal responses were registered using a piezo-electric sensor mounted under the lower jaw. b Corticosterone (CORT) increase in response to the adrenocorticotropic hormone (ACTH) stimulation test (n = 3). Mean CORT concentration increases over time after ACTH administration. c Absolute change in plasma corticosterone on habituation days D1, D4, and D7. Absolute change in CORT was significantly lower on habituation day 4 and day 7 compared to habituation day 1. Bars = mean values of absolute corticosterone level change (n = 4). **p = 0.016, *p = 0.037 (one-way repeated-measure ANOVA, a Sidak correction for multiple comparison). d Heart rate during the habituation procedure on days 1, 4, and 7 (n = 4). Heart rate was significantly lower on habituation day 7 compared to habituation day 1 (dotted line = baseline at rest). *p = 0.023 (one-way repeated-measure ANOVA, a Sidak correction for multiple comparison). e, f Median absolute deviation (MAD) of estimated motion parameters (translation and rotation) over resting-state and active decision task (n = 8). g Example of activation clusters caused by 2 s visual stimulation (calculated for an exemplary pigeon brain for the first run) to estimate the hemodynamic response function parameters. The activation mask from individual pigeons was applied to the other runs to extract the time course of hemodynamic responses within each ROI. h Left curve shows the BOLD signals (HDR) for different visual stimuli. Data are represented as mean ± SEM (n = 5 animals, number of trials per condition is 300). The right curve shows the best-fitted pigeon hemodynamic response function (α₁ = 7.71, α₂ = 11.48, β₁ = 1.74, β₂ = 0.74, c = 0.25). i Estimated parameters of HDR for different colors. Hemodynamic responses did not differ between colors. Each circle in b–f and i represent a single value per individual during different experiments. ACTH adrenocorticotropic hormone, CORT corticosterone, BOLD blood-oxygen-level-dependent, H height, T time-to-peak, W full-width at half-maximum, A anterior, P posterior, L left, R right. Source data are provided as a Source Data file.

Fig. 2. Behavioral results of the color discrimination task.

a Top row shows the experimental design. All events were equiprobable. Bottom row shows the timing of an individual Go (S+) trial. Animals were rewarded when mandibulating during S+ presentations. To avoid stimulus intensity effects during training, two different light intensities (20 and 100% of maximum possible intensity) were used. b Raster plot of a typical session from one pigeon. Time increase from bottom to top for the session and left to right represents the event within single trials (with time-point 0 corresponding to trial onset). During the depicted session, red-light signaled S+ and green-light S− trials. Mandibulation was tracked during each single trial. Blue dots represent the animal’s mandibulations, which happened mostly during S+ presentations and reward periods. c Histogram showing an overlay of mandibulations both during Go and NoGo trials in a group analysis of eight pigeons (with time-point 0 corresponding to stimulus onset). Dotted line shows the end of the post-reward period. d Mandibulation rate histogram. Group analysis showed that mandibulation rate over all S+ trials was significantly higher than over all S− trials (two-tailed t test, t(654) = 9.66, p = 9.99e − 21). e Behavioral results of all pigeons over all test sessions. Each data point represents the projected Hit/FA pairs to the ROC space. Source data are provided as a Source Data file.

Fig. 3. BOLD response pattern during the color discrimination task.

Coronal images at different levels of the pigeon telencephalon with overlay of anatomical borders based on the pigeon MRI atlas. Pigeon atlas images are shown in gray scale, while the areas activated significantly during the color discrimination task (contrast of Go > NoGo + mandibulation) are highlighted in color (Z = 3.1 and p < 0.05 FWE corrected at the cluster level, group analysis). Results demonstrate activated clusters in a visual, olfactory, auditory, trigeminal, tactile, and polysensory areas; limbic and basal ganglia components; and b memory- and executive control-related pallial areas (see also Fig. 5 and Supplementary Fig. 6). The last image on the left column shows a 3D representation of the pigeon brain with an example window at level A 8.50. Anatomical borders in b are based on study by Herold et al. and color-coded areas (red, green, and yellow) represent the terminal fields of afferents from secondary sensory areas on the NCL, based on Kröner and Güntürkün. The last image of the right column shows a 3D representation of the selected coronal images from NCL. HI high intensity, LH low intensity. The corresponding abbreviations of ROIs are listed in the Supplementary Table 3.

Fig. 4. Inhibitory and mandibulation response patterns.

a Statistical activation maps for signal increase in the contrast of NoGo > Go (Z = 3.1 and p < 0.05 FWE corrected at the cluster level, group analysis). b 3D representation of statistical activation map for signal increase from the contrast of mandibulation > rest (Z = 3.1 and p < 0.05 FWE corrected at the cluster level, group analysis). The activation maps were registered and illustrated on the pigeon atlas. The activation significance is demonstrated by the color scale. HbL lateral habenula, SpA subpallialis amygdalae, Bas nucleus basalis, NFL nidopallium frontolaterale.

Fig. 5. Schematic depiction of the telencephalic network in the pigeon brain active during our color discrimination experiment.

Nodes represent active regions in the “Go > NoGo + mandibulation” contrast (for more details see Fig. 3 and Supplementary Fig. 6). Nodes are colored based on their involvement in different functional networks. Note that the activated networks are spatially distributed rather than localized. For abbreviations see Supplementary Table 3.