Jaan Aru

Untangling cross-frequency coupling in neuroscience.

Cross-frequency coupling (CFC) has been proposed to coordinate neural dynamics across spatial and temporal scales. Despite its potential relevance for understanding healthy and pathological brain function, the standard CFC analysis and physiological interpretation come with fundamental problems. For example, apparent CFC can appear because of spectral correlations due to common non-stationarities that may arise in the total absence of interactions between neural frequency components. To provide a road map towards an improved mechanistic understanding of CFC, we organize the available and potential novel statistical/modeling approaches according to their biophysical interpretability. While we do not provide solutions for all the problems described, we provide a list of practical recommendations to avoid common errors and to enhance the interpretability of CFC analysis.

The Phase of Thalamic Alpha Activity Modulates Cortical Gamma-Band Activity: Evidence from Resting-State MEG Recordings

Recent findings have implicated thalamic alpha oscillations in the phasic modulation of cortical activity. However, the precise relationship between thalamic alpha oscillations and neocortical activity remains unclear. Here we show in a large sample of healthy human participants (n = 45) using spatial filtering techniques and measures of phase amplitude coupling that the amplitude of gamma-band activity in posterior medial parietal cortex is modulated by the phase of thalamic alpha oscillations during eyes-closed resting-state recordings. In addition, our findings show that gamma-band activity in visual cortex was not modulated by thalamic alpha oscillations but coupled to the phase of strong cortical alpha activity. To overcome the limitations of electromagnetic source localization we estimated conduction delays using transfer entropy and found nonspurious information transfer from thalamus to cortex. The present findings provide novel evidence for magneto-encephalography-measured phase coupling between cortical gamma-band activity and thalamic alpha oscillations, which highlight the role of phasic inhibition in the coordination of cortical activity.

Local Category-Specific Gamma Band Responses in the Visual Cortex Do Not Reflect Conscious Perception

Which neural processes underlie our conscious experience? One theoretical view argues that the neural correlates of consciousness (NCC) reside in local activity in sensory cortices. Accordingly, local category-specific gamma band responses in visual cortex correlate with conscious perception. However, as most studies manipulated conscious perception by altering the amount of sensory evidence, it is possible that they reflect prerequisites or consequences of consciousness rather than the actual NCC. Here we directly address this issue by developing a new experimental paradigm in which conscious perception is modulated either by sensory evidence or by previous exposure of the images while recording intracranial EEG from the higher-order visual cortex of human epilepsy patients. A clear prediction is that neural processes directly reflecting conscious perception should be present regardless of how it comes about. In contrast, we observed that although subjective reports were modulated both by sensory evidence and by previous exposure, gamma band responses solely reflected sensory evidence. This result contradicts the proposal that local gamma band responses in the higher-order visual cortex reflect conscious perception.

Multiagent cooperation and competition with deep reinforcement learning

Evolution of cooperation and competition can appear when multiple adaptive agents share a biological, social, or technological niche. In the present work we study how cooperation and competition emerge between autonomous agents that learn by reinforcement while using only their raw visual input as the state representation. In particular, we extend the Deep Q-Learning framework to multiagent environments to investigate the interaction between two learning agents in the well-known video game Pong. By manipulating the classical rewarding scheme of Pong we show how competitive and collaborative behaviors emerge. We also describe the progression from competitive to collaborative behavior when the incentive to cooperate is increased. Finally we show how learning by playing against another adaptive agent, instead of against a hard-wired algorithm, results in more robust strategies. The present work shows that Deep Q-Networks can become a useful tool for studying decentralized learning of multiagent systems coping with high-dimensional environments.

Phenomenal awareness can emerge without attention.

In a recent debate, the views that top-down attention is necessary for consciousness (Cohen et al., 2012a,b) and that consciousness is independent of top-down attention (Tsuchiya et al., 2012) have clashed. Here, we list the overlooked or ignored arguments that should be considered before deciding that consciousness is inevitably the result of attention. The issue of relation of bottom-up attention and consciousness seems to have two possible solutions according to the evidence available at present. First, based on the current evidence it may be agreed upon that conscious experience cannot be dissociated from bottom-up exogenous attention (Tsuchiya et al., 2012). Second, it is possible that there is phenomenal consciousness emerging without bottom-up attention, but empirical evidence for this is either lacking or too much controversial at present. Here, we consider some instances where simple conscious experience, i.e., phenomenal awareness emerges without top-down attentional deployment: Sensory experiences can be brought about artificially by brain stimulation such as when using transcranial magnetic stimulation (TMS). Visual scotomas (Murd et al., 2010) or phosphenes as “visual echoes” (Jolij and Lamme, 2010) generated by early visual cortex stimulation serve as an example for simple conscious experiences evoked by TMS. Attention is not needed for these conscious experiences. Such artificial experience can be evoked at an unexpected moment and location and will become available for attention only subsequently. It could be argued that in the task where subjects evaluate their phosphenes or scotomas the subjects are still attending to these conscious phenomena. However, we claim that if a subject is performing an attending task unrelated to TMS and visual perception, but is unexpectedly given a single TMS pulse which elicits a phosphene, this phosphene will be consciously perceived. Similarly, involuntary hallucinations and pathological sensations emerge without any attention to these contents. For example consider tinnitus or auditory verbal hallucinations. It would be a simple remedy for the patients if simply not attending to the hallucinations would make them disappear. Many such hallucinations come and go without the subject having any control of their duration or onset (Sacks, 2012). No attention is required for dream episodes. Dreams are conscious experiences which are generally characterized by the lack of top-down attention (Hobson, 2002). The loss of top-down control and attention in typical everyday dreams is also evidenced by the peculiar feeling that accompanies lucid dreaming, where due to training some top-down control over the dream content becomes available for the subject (e.g., Voss et al., 2009). Experiments and phenomenology on microgenetic formation of perceptual images and visual immediate (iconic) memory demonstrate that phenomenal-perceptual proto-objects or scenes can precede attentional selection (Bachmann, 2000; Lamme, 2004, 2010) and that non-attended objects are consciously experienced (Vandenbroucke et al., 2012). These arguments show that simple conscious experience could emerge independently of attention. One could argue that in all these cases even when the conscious experience is very simple (as in the case of phosphenes) the stimulus is first represented in the brain, then amplified by top-down attention and only then emerges in awareness. Thus, one could claim that these arguments are not sufficient for showing that simple conscious experience can emerge without attention. However, even if one would agree that the evidence does not convincingly demonstrate the independence of attention and consciousness, it would not imply that attention is necessary for consciousness. Furthermore, there are some additional arguments to support the independence of attention and consciousness. If attention is necessary for conscious perception, attending to certain phenomenal content should not be unfavorable for experiencing this content. However, several recent studies have shown that attention has a detrimental effect on the duration of afterimages or sensory aftereffects (Lou, 2001; Bachmann and Murd, 2010; van Boxtel et al., 2010). If attention readily eliminates phenomenal experience then the awareness-related mechanisms whose activity is suppressed together with phenomenal experience must be independent from the attentional mechanisms. Attending to the contents of the currently dominating image in binocular rivalry does not preclude that the competing alternative might spontaneously emerge in phenomenal awareness. If attention is necessary for consciousness and it is deployed to the dominating image, how could the suppressed image access awareness? Many authors agree that consciousness and attention have different neurobiological mechanisms (Lamme, 2004; Koch and Tsuchiya, 2007; Bachmann, 2011). This of course does not deny interaction between these mechanisms in typical cases where attention has a supportive role in bringing perceptual contents to consciousness, but the independence of the neurobiological mechanisms supports the view that attention and consciousness are autonomous processes. Finally, it is not true that in order to support the independence of consciousness and attention one has to show that attention is not needed for the stimulus to enter consciousness across a set of experimental paradigms (Cohen et al., 2012a). In fact, one could turn the argument around: if it is consistently shown by many research groups that attention is not needed for conscious perception even in one single paradigm, then this already implies that attention is not necessary for consciousness. It was also argued that the absence of the attentional effect on consciousness is a null finding which should be treated with caution (Cohen et al., 2012a). However, if this null finding is consistent over several experiments with sufficient statistical power, this result is solid by any scientific standard. Taken together, these arguments support the view that attention and consciousness are independent from each other, but can interact. Phenomenal consciousness can emerge without attention.

Does the P300 reflect conscious perception or its consequences

A common scientific strategy toward understanding consciousness is to study neural correlates of consciousness (NCC) for a particular conscious percept. This can be done by contrasting conditions in which subjects are aware and unaware of a particular visual stimulus. However, recent findings have been contradictory and this approach appears not to reveal only the NCC, but also the prerequisites or consequences of consciousness. The goal of the present study was to investigate whether the P300 component often claimed to be a key signature of conscious access might actually rather reflect the consequences of conscious perception. Subjects participated in an electroencephalography (EEG) experiment where most of the stimuli were clearly perceived and only a quarter of trials was associated with not perceiving the target. The visual masking paradigm used no discrimination task, always the same stimulus was presented and an additional task was given together with the target stimulus. Results indicate that trials where subjects reported to have seen the stimulus are associated with a more pronounced P300. Hence, the present data support the theories which claim that P300 is a marker of conscious perception. However, an earlier component, visual awareness negativity also tracks conscious perception reliably so that the P300 need not be the earliest correlate of conscious perception.

Caffeine enhances frontal relative negativity of slow brain potentials in a task-free experimental setup.

State dependent effects on brain processes are difficult to study due to the task-related confounds. Even in simple task environments external stimuli inevitably interact with dynamically changing states of the brain. Psychopharmacological manipulation and transcranial magnetic stimulation can be used independently of variations in subjects experimental task and environmental stimulation. Our aim was to show the investigative potential of combining these two methods for studying the effects of the state of the brain on the dynamics of task-free evoked brain activity. Caffeine was used for inducing higher arousal state and transcranial magnetic stimulation was used to evoke widespread bioelectrical responses of the brain. Occipitally delivered magnetic pulses caused increased global negativity of the brain potentials, but no speed-up of brain potentials when caffeine was administered. The relative negativization effect was most clearly expressed in slow potentials and as measured from frontal and parietal electrodes. This study shows how the causal effects of brain states on neural processes can be studied without the confounding influence of experimental task and stimuli.

Early effects of previous experience on conscious perception

Abstract Constructive theories of brain function such as predictive coding posit that prior knowledge affects our experience of the world quickly and directly. However, it is yet unknown how swiftly prior knowledge impacts the neural processes giving rise to conscious experience. Here we used an experimental paradigm where prior knowledge augmented perception and measured the timing of this effect with magnetoencephalography (MEG). By correlating the perceptual benefits of prior knowledge with the MEG activity, we found that prior knowledge took effect in the time-window 80–95 ms after stimulus onset, thus reflecting an early influence on conscious perception. The sources of this effect were localized to occipital and posterior parietal regions. These results are in line with the predictive coding framework.

General Markers of Conscious Visual Perception and Their Timing

Previous studies have observed different onset times for the neural markers of conscious perception. This variability could be attributed to procedural differences between studies. Here we show that the onset times for the markers of conscious visual perception can strongly vary even within a single study. A heterogeneous stimulus set was presented at threshold contrast. Trials with and without conscious perception were contrasted on 100 balanced subsets of the data. Importantly, the 100 subsets with heterogeneous stimuli did not differ in stimulus content, but only with regard to specific trials used. This approach enabled us to study general markers of conscious visual perception independent of stimulus content, characterize their onset and its variability within one study. N200 and P300 were the two reliable markers of conscious visual perception common to all perceived stimuli and absent for all non-perceived stimuli. The estimated mean onset latency for both markers was shortly after 200 ms. However, the onset latency of these markers was associated with considerable variability depending on which subsets of the data were considered. We show that it is first and foremost the amplitude fluctuation in the condition without conscious perception that explains the observed variability in onset latencies of the markers of conscious visual perception.

Right-frontal slow negative potentials evoked by occipital TMS are reduced in NREM sleep

Occipital transcranial magnetic stimulation applied in a task-free experimental setup leads to enhanced relative negativity of frontally recorded evoked slow potentials under the influence of caffeine (Murd et al., 2010 [26]). We tested whether this increased negativity could be reversed when a similar magnetic stimulation is applied during quiet sleep where consciousness is absent. Consistently with the hypothesis, non-REM sleep led to relative more positive slow brain potentials, compared to wakefulness. This effect was lateralized to the right hemisphere. We conclude that TMS indeed elicits slow negative potentials in higher arousal states, but the effect has hemispheric specificity depending on how arousal is manipulated.