By Thomas Pfeffer and Tobias Donner.
The catecholaminergic neurotransmitters noradrenaline and dopamine are important regulators of brain state, and their function is impaired in several psychiatric disorders. An influential account posits that noradrenaline renders inference and decision-making more variable (exploratory) when faced with unknown environmental contingencies (Aston-Jones & Cohen, 2005).
We tested this idea by means of an integrative approach in the human brain. We quantified if and how catecholamines alter the variability of perceptual inference and of the underlying neural population activity in the cerebral cortex. In three experimental sessions, we selectively boosted catecholamine levels (using the noradrenaline transporter inhibitor atomoxetine) or acetylcholine levels (using the cholinesterase inhibitor donepezil), or administered a placebo. Both drugs are routinely used in the clinical practice. In each session, we measured cortical population activity using magnetoencephalography (MEG) during two behavioral contexts (Figure 1A): (i) eyes open “resting-state”, during which participants simply fixated a green dot in the center of an otherwise gray screen (henceforth referred to as “Fixation”); and (ii) perception of ambiguous visual input. In the latter context, participants watched a seemingly rotating sphere (see embedded video below), which induced ongoing, spontaneous alternations in the perceived rotation direction. Previous work in monkeys and humans has shown that motion-sensitive visual cortical regions in the occipital and parietal lobes encode the perceived direction (Parker et al, 2002; Brouwer & van Ee, 2007). Participants either reported each alternation immediately by pressing a button, or silently counted the number of alternations and reported the number after each block of 10 min. The latter condition, referred to as “Task-counting”, was similar to “Fixation” in that it did not entail any changes in the sensory input or motor responses during the MEG recording. Both conditions enabled us to analyze intrinsic fluctuations of cortical population activity, as a function of pharmacological condition and behavioral context.
Catecholamines, but not acetylcholine, consistently increased the number of perceptual alternations (Figure 1B). In other words, catecholamines boost the variability of the perceptual interpretation of constant, but ambiguous sensory input. Previous theoretical and experimental work (Moreno-Bote et al, 2007; Van Loon et al, 2013) indicates that such an effect can be mediated by an increase in the ratio between excitation and inhibition (E/I ratio) in the cortical regions that process the stimulus.
Another marker of cortical E/I ratio based on the MEG measurement pointed to the same conclusion. Catecholamines, but not acetylcholine, altered the variability of band-limited MEG activity in occipital and parietal cortex. Specifically, catecholamines increased the so-called long-range temporal correlations (LRTCs) of spontaneous fluctuations in the amplitude of alpha-band oscillations (Figure 1C), a widely observed signature of intrinsic cortical activity that is attracting substantial interest as a potential marker of ‘self-organized criticality’ (Palva et al, 2013; Singer, 2013). Simulations of an integrate-and fire neural network revealed that subtle changes in E/I ratio through synaptic gain modulation changed these same LRTCs in the model (Figure 1D). However, the change in LRTCs as a function E/I ratio was non-monotonic (inverted U-shape). Thus, the direction of the change in E/I ratio from a change in LRTCs can only be if the baseline E/I ratio is known, which is generally not the case (Figure 1E, gray dots). Fortunately, for our current experimental setting, the baseline E/I ration during Task-counting (strong visual drive) was constrained through recent insights from invasive physiology in rodents: visual stimulation pushes visual cortex into an inhibition-dominant regime (Haider et al, 2013; Adesnik et al, 2017). Assuming generalization of this mechanism to human cortex (Figure 1E,F, yellow dot), we could infer the direction of change in E/I ratio from the observed increase in LRTCs due to catecholamines: an increase in E/I ratio (Figure 1F, pink dot), consistent with the reported behavioral effects.
In sum, catecholamines alter the variability of perceptual inference as well as a specific feature of intrinsic cortical dynamics (LRTCs). Mechanistically, both effects can be explained by an increase in E/I ratio within the cortical circuits processing the visual stimulus. It will now be important to test if the catecholamine-induced changes in intrinsic variability in brain and cognition are adaptive, that is, are specifically instigated in the face of uncertainty or ambiguity about the state of the environment or the best course of action (Aston-Jones & Cohen, 2005).
Because our approach allows for noninvasively inferring changes in E/I ratio in humans, it also opens up exciting translational perspective. Different from other recently proposed, model-based behavioral readouts of E/I-ratio (e.g. Lam et al, 2017; Jardri et al, 2017), monitoring the spontaneous perceptual alternations induced by an ambiguous stimulus is not challenging. Similarly, fluctuations in band-limited activity can be easily measured using widely available EEG systems. Hence, our approach lends itself to application in clinical populations associated with E/I imbalances (e.g., schizophrenia or autism) or to the assessment of the effects of drugs acting on this balance.
Reference: Pfeffer T, Avramiea A-E, Nolte G, Engel AK, Linkenkaer-Hansen K, Donner TH (2018) Catecholamines alter the intrinsic variability of cortical population activity and perception. PLoS Biol 16(2): e2003453. https://doi.org/10.1371/journal.pbio.2003453