By Tobias H. Donner.
I just got back from the 2019 Gordon Research Conference on Eye Movements Bridging Knowledge About Gaze Control Across Applications, Disorders, Levels and Species. I presented in the session Pupillometry as a Behavioral Indicator, along with Jake Reimer (Baylor College), Sidd Joshi (U Penn), and Doug Munoz (Queen’s U), and chaired by Brian Corneil (Western U) – an awesome group of fellow speakers. This appears to have been the first pupillometry session at GRC Eye Movements. We all tried hard for it not to be the last one!
Brian opened the session with a nice historical perspective on early research into pupil motility as a read-out of the peripheral autonomic nervous system (its sympathetic and parasympathetic arms) and the associated central nuclei of the brainstem. He started from a description of the pupil light response (i.e., constriction), which is nowadays used in every neurological examination to assess the integrity of the cranial nerves and brainstem centers. He then reviewed some of the early work linking non-luminance mediated pupil dilation to cognition, ranging from 1960’s psychology studies of arousal and effort to the highly influential 2005 review paper by Gary Aston-Jones and Jonathan Cohen (2005). He finally discussed the relationship between pupil dilation and the orienting response. Already in his pioneering work on the functional localization of the cerebral cortex in late 19th century, David Ferrier had found the pupil to dilate when stimulating the vicinity of the frontal eye fields in monkeys. Brian’s lab has recently confirmed this effect in a controlled and quantitative microstimulation experiment, specifically for sub-saccadic stimulation levels (Lehmann & Corneil, 2016).
Jake reviewed his insights into the neural underpinnings of pupil-linked cortical state fluctuations in awake mice. He showed that many established neural signatures of the “attentive state” (for example reduction of trial-by-trial variability of sensory responses) are evident in V1 during strong pupil dilation (Reimer et al, 2015). Imaging noradrenergic axons from the brainstem locus coeruleus (LC) and cholinergic axons from the basal forebrain to cortex shows correlations between both neuromodulatory systems and pupil dilation (Reimer, McGinley, et al, 2016). In the final part of the talk, he presented the exciting new techniques that can now be used in the mouse model to advance the mechanistic understanding of cortical state. For example, cellular resolution multi-area imaging, imaging subthreshold activity with voltage sensors, and simultaneous imaging of anatomically connected neurons within an area will be instrumental for understanding how arousal shapes cortical information flow.
Sidd began by reviewing his monkey physiology work showing several tight links between pupil dilation and neural activity in the LC – during slow, ongoing fluctuations in LC activity, stimulus-evoked responses (startling tones), as well as direct, electrical LC-stimulation (Joshi et al, 2016). Importantly, however, similar associations hold also for a number of further brain regions analyzed in this study, in particular, the inferior and superior colliculi (IC and SC) or the anterior cingulate cortex (ACC). In other words, while this work (as Jake’s) clearly establishes the LC as one important source of non-luminance mediated variations in pupil diameter, the central control of pupil-linked arousal appears, like most functions, complex and distributed across many different brain regions. Sidd closed with exciting new data from simultaneous recordings of neural activity in monkey LC and ACC during pupillometry. Those measurements indicate, among other things, that LC and pupil linked changes in ACC correlations depend on the nature of LC firing. During passive fixation and tonic LC firing, ACC neurons were decorrelated. But when driven by startling sounds that drove phasic increases in LC firing, ACC correlations increased. Thus, the nature of LC firing can potentially influence network activity in ACC: decorrelation is generally considered to improve signal-to-noise ratios and fidelity of information transmission while synchronization can enhance temporal alignment of activity in response to a sudden change in the environment.
I put the focus on the function of rapid (“phasic”) boosts in central arousal, which occur during challenging decisions and are – not surprisingly – associated with pupil dilation. I briefly reiterated the link between such phasic pupil dilations and LC-responses, now focusing on the context of behavioral choice tasks. I then reviewed our recent work into the functional role of these phasic arousal boosts for decision-making (e.g. de Gee et al, 2014, 2017, 2018; Urai et al, 2017; Colizoli et al, 2018). The common finding is that large phasic arousal responses predict a reduced impact of intrinsic biases (or prior expectations) on behavioral choice. This relationship holds for humans and mice, across various domains of decision-making – from low-level perceptual decisions to higher-level, numerical or memory-based decisions. In all cases, the “locus” of the arousal-linked bias suppression in the decision process lies specifically in the gradual accumulation of decision evidence over time, which underlies many decisions (de Gee et al, 2017, 2018). I closed with new data tapping into the potential normative function of these dynamic bias modulations in real-life decision-making: during evidence accumulation in environments that can undergo hidden changes, evidence samples strongly indicative of a change drive strong arousal responses. This arousal boost, in turn, is associated with increased leverage of the corresponding evidence on the final choice – in other words, reducing the impact of the existing belief in the evidence accumulation process and promoting belief change (Murphy et al, SfN Abstracts, 2019).
Jake, Sid and I focused on ascending neuromodulatory systems (LC-noradrenaline and basal forebrain-acetylcholine systems) as the candidate sources of pupil-linked variations in brain state. In the final talk of the session, Doug provided an alternative perspective on the neural basis of task-evoked pupil size modulations and the associated changes in brain state. He reviewed work from his lab showing that a pathway involved in the orienting response can also drive pupil dilation. The SC lies at the center of this pathway (Wang et al, 2012, 2015, 2015). This pathway is rapid and presumably inhibits the Edinger-Westphal nucleus that, in turn, controls pupil constriction. Since SC is reciprocally connected with a large number of cortical regions, this pathway is an alternative, presumably faster, route by which cognitive signals from specific cortical regions can drive coordinated state changes across widespread brain networks. Doug developed a large-scale circuit diagram that presents this SC-pathway next to the LC-pathway the previous speakers focused on. Clearly, future work is needed to dissect these pathways and their functions. Doug closed the session by highlighting the potential of pupil responses as biomarkers for a variety of neuropsychiatric disorders, such as Parkinson’s or Alzheimer’s disease. Indeed, his lab has already shown that the pupil light response is altered in a wide range of such disorders – in other words, it provides a simple signal for the detection of alterations in brain function. This raises the intriguing possibility that pupil responses in cognitive tasks might be used for discriminating between different disorders.
Taken together, this was a highly integrative session on a nascent and still relatively small sub-field of systems neuroscience. I think, not surprisingly, that this field has enormous potential for advancing both basic and translational neuroscience. It is a tantalizing perspective that psychiatrists may one day be able to infer specific aberrations of brainstem-cortex interactions from pupil responses during well-controlled cognitive tasks – in much the same way as neurologists have been inferring specific lesions in cranial nerves and brainstem nuclei from alterations in the pupil light response.
Clearly, reaching this point requires us to make much headway through a landscape full of uncertainty. But this way has been paved by the insights and technical approaches featured in this symposium, as well as the work of a growing number of other scientists who have begun to study the relationship between pupil responses, brain state, and cognition. GRC Eye Movements could be a wonderful venue for featuring this field in the future. In the meantime, keep your pupils widely dilated for taking in the latest developments in this exciting research field.
Thanks to all speakers as well as Anne Urai, Jan Willem de Gee, and Peter Murphy for their feedback on an earlier draft of this post.
Further reading:
Task-evoked pupil responses reflect internal belief states. Colizoli O, de Gee JW, Urai AE, Donner TH. Sci Rep. 2018 Sep 12;8(1):13702.
An integrative theory of locus coeruleus-norepinephrine function: adaptive gain and optimal performance. Aston-Jones G, Cohen JD. Annu Rev Neurosci. 2005;28:403-50.
Phasic arousal suppresses suboptimal decision biases in mice and humans. de Gee JW, Tsetsos K, Schwabe L, Urai AE, McCormick DA, McGinley MJ, Donner TH. bioRXiv 2018: 447656.
Decision-related pupil dilation reflects upcoming choice and individual bias.Dynamic modulation of decision biases by brainstem arousal systems. de Gee JW, Colizoli O, Kloosterman NA, Knapen T, Nieuwenhuis S, Donner TH. Elife. 2017 Apr 11;6. pii: e23232.
Decision-related pupil dilation reflects upcoming choice and individual bias. de Gee JW, Knapen T, Donner TH. Proc Natl Acad Sci U S A. 2014 Feb 4;111(5):E618-25.
Relationships between Pupil Diameter and Neuronal Activity in the Locus Coeruleus, Colliculi, and Cingulate Cortex. Joshi S, Li Y, Kalwani RM, Gold JI. Neuron. 2016 Jan 6;89(1):221-34.
Transient Pupil Dilation after Subsaccadic Microstimulation of Primate Frontal Eye Fields. Lehmann SJ, Corneil BD. J Neurosci. 2016 Mar 30;36(13):3765-76.
Waking State: Rapid Variations Modulate Neural and Behavioral Responses. McGinley MJ, Vinck M, Reimer J, Batista-Brito R, Zagha E, Cadwell CR, Tolias AS, Cardin JA, McCormick DA. Neuron. 2015 Sep 23;87(6):1143-1161.
Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Reimer J, McGinley MJ, Liu Y, Rodenkirch C, Wang Q, McCormick DA, Tolias AS. Nat Commun. 2016 Nov 8;7:13289.
Pupil fluctuations track fast switching of cortical states during quiet wakefulness. Reimer J, Froudarakis E, Cadwell CR, Yatsenko D, Denfield GH, Tolias AS. Neuron. 2014 Oct 22;84(2):355-62.
Pupil-linked arousal is driven by decision uncertainty and alters serial choice bias. Urai AE, Braun A, Donner TH. Nat Commun. 2017 Mar 3;8:14637.
Multisensory integration in orienting behavior: Pupil size, microsaccades, and saccades. Wang CA, Blohm G, Huang J, Boehnke SE, Munoz DP. Biol Psychol. 2017 Oct;129:36-44.
A circuit for pupil orienting responses: implications for cognitive modulation of pupil size. Wang CA, Munoz DP. Curr Opin Neurobiol. 2015 Aug;33:134-40.
Microstimulation of the monkey superior colliculus induces pupil dilation without evoking saccades. Wang CA, Boehnke SE, White BJ, Munoz DP. J Neurosci. 2012 Mar 14;32(11):3629-36.