Neural basis of perceptual decisions
Social and reward influences
Neurophysiology, Imaging, Psychophysics, Histology
Visual Perception and Decision-Making
neural circuits For Perception aND Decision-Making In Primates
My research group seeks to explain and alter perceptual decision-making from the level of single brain cells through to mental states. With this work, we aim to understand the neuronal code underlying conscious processes. One fundamental problem is that neuronal activity sometimes represents processes of which we are aware and sometimes codes for information to which we have no access (Krug et al. 2004). Using electrical microstimulation of neurons in Rhesus monkeys, we can show how the activity of neurons in visual cortex causally contributes to the perceptual appearance of visual objects. For instance, we have identified a strong cognitive signal in the activity of single neurons in extrastriate visual area V5/MT that shapes perceptual decisions about 3D-motion figures (Dodd et al. 2001; Krug et al. 2013). This brain area in Rhesus monkeys has a structural and functional homologue in humans (Large et al. 2016). We are currently investigating how contextual effects, like expected reward and social influence, interact with sensory signals in the brain and thus affect visual perception. This has profound implications for our understanding of decision-making in healthy individuals and in individuals with a psychiatric disorder.
Comparing structural MRI and histology in individual brain
A direct comparison of myelin signals in a specific block of primate tissue with in vivo structural MRI, post mortem structural MRI and histological staining. We show that structural MRI in vivo accurately detects myelin-related signals in primate cortex. Large et al (2016) Cerebral Cortex 26 (10) cover.
Identifying cortical areas with myelin in vivo
A map of cortical myelin (blue-red scale) for a macaque monkey obtained in vivo. Areas with strong myelin signal (red) are for example V1, V5/MT, MST and M1.
MRI image of a Manganese injection in V5/MT (1h)
Visual motion and depth discrimination are pertinent models to study the brain signals for perception and decision-making. A number of different visual cortical areas have been implicated in the perception of motion and 3D depth. Brain areas have traditionally been studied in isolation from one another; and yet a network of areas appears to be required. The challenge has been to determine what visual cortical components are involved and how they are interconnected. The Krug Lab is using a multifaceted approach to reveal the anatomical connections that come and go from the middle temporal area V5/MT in primates - an area with a central role in the perception of motion and 3D depth. By comparing brain connectivity visualized with cutting egde non-invasive MRI methods (e.g. Diffusion-Weighted Imaging, resting-state fMRI), invasive MRI methods (e.g. Manganese-tracing) as well as histological tracing in the same individual, we also provide validation of these new imaging techniques.
Characterizing neuronal circuits in V5/MT for perceptual decisions about 3D and visual motion
Neurons in extrastriate visual area V5/MT are tuned to conjunctions of binocular 3D depth and direction of motion. Based on these properties, their signals correlate with perceptual switches in bistable structure-from-motion figures (Dodd et al. 2001). Neurons with the same stimulus selectivity are clustered (Albright et al 1984; DeAngelis & Newsome 1999). Different studies, psychophysical and neuronal, have postulated that specific intrinsic connectivity in V5/MT underlies the emergence of such signals (Bradley et al., 1998; Nawrot & Blake 1991). We have shown that long range connectivity in V5/MT is highly patterned, consistent with specific connectivity between functionally distinct depth and direction columns (Ahmed et al 2012). We are currently investigating the nature and specificity of neuronal connections within and between visual areas in relation to their functional properties.
Picture above shows a cortical V5/MT injection of CTb labeling a column of neurons (Ahmed et al 2012)
Expected reward changes visual processing
Changing Perception through electrical microstimulation in cortical area V5/MT
Judgments about the perceptual appearance of visual objects require the combination of multiple parameters, like location, direction, color, speed, and depth. Our understanding of perceptual judgments has been greatly informed by studies of ambiguous figures, which take on different appearances depending upon the brain state of the observer. Resolving the rotation direction of ambiguous structure-from-motion (SFM) cylinders requires the conjoint decoding of direction of motion and binocular depth signals. Within cortical visual area V5/MT of two macaque monkeys, we applied electrical stimulation at sites with consistent multiunit tuning to combinations of binocular depth and direction of motion, while the monkey made perceptual decisions about the rotation of SFM stimuli. For both ambiguous and unambiguous SFM figures, rotation judgments shifted as if we had added a specific conjunction of disparity and motion signals to the stimulus elements. Krug et al (2013) is the first causal demonstration that the activity of neurons in V5/MT contributes directly to the perception of SFM stimuli and by implication to decoding the specific conjunction of disparity and motion, the two different visual cues whose combination drives the perceptual judgment.
Micro-pool of decision neurons may include less than perfect responses
In the primate visual cortex, neurons signal differences in the appearance of objects with high precision. However, not all activated neurons contribute directly to perception. We defined the perceptual pool in extrastriate visual area V5/MT for a stereo-motion task, based on trial-by-trial co-variation between perceptual decisions and neuronal firing. We manipulated the activity of single neurons trial-to-trial by introducing task-irrelevant stimulus changes. For individual neurons, the correlation between perceptual choice and neuronal activity may be fundamentally different when responding to different versions of the same stimulus in the same location. Therefore, neuronal pools supporting sensory discrimination must be structured flexibly and independently for each stimulus configuration to be discriminated. Krug et al 2016 is part of the themed Phil Trans B issue 'Vision in our three-dimensional world'.
Development of social influence in children with autism
The opinions of others have a profound influence on human decision-making. Graduate student Imogen Large tested 125 neurotypical and 30 autistic children between 6 and 14 years on a visual stereo-motion task under social influence. In neurotypical children, systematic bias in favour of advice provided by another person emerged around 12 years of age. Drift-diffusion modelling indicated that this social bias was best explained by changes in sensory processing of the visual stimulus that children were judging, rather than a change in decision behavior. By contrast, children with autism did not develop a perceptual bias by the same age. We conclude that by the early teens, typical neurodevelopment allows social influence to systematically bias perceptual processes - previously linked to dorsal stream visual areas for this task. That the same bias does not emerge in autistic children could potentially explain some of their difficulties when responding to social cues.
Graphs above show drift diffusion rates (y-axis) for different cylinder stimuli (along x-axes) and under different advice conditions (green + blue; advice 'cylinder rotates right'; advice 'cylinder rotates left').
Human brain signals in real time
Postdoc Nela Cicmil recorded MEG brain responses when human subjects made a decision about the appearance of a rotating cylinder. Early brain activation is found at 50-100ms in primary visual cortex (top), progressing to hMT+ at 150-200 msec reaching parietal cortex around 250 msec.