The 3rd Annual Meeting of the Neurophilosophy of Free Will Consortium
March 11th~14th, Palm Springs, CA
Kimpton Rowan Palm Springs Hotel
To request a shuttle pickup from the Palm Springs Airport (PSP), please call the hotel front desk at (760) 904-5015.
The meeting check-in/registration desk is outside the Sky Ballroom (the main meeting space), located on the 2nd floor of the hotel.
Please come by the desk and pick up your name badge between 1PM and 5PM on Friday, March 11th.
Any meeting-related question can be directed to Tian Lan at firstname.lastname@example.org.
Meeting Registration/Check-in – 2nd floor of the hotel
Degrees of Freedom: A Discussion
Freedom (to do or will a particular act) is supposed to vary along five dimensions:
- (A) How many barriers is the agent free from?
- (B) How many alternatives is the agent free to do?
- (C) How many reasons for or against the act is the agent able to respond to?
- (D) How conscious was the agent of what was being done?
- (E) How intentional or pre-planned was the act?
Participants will discuss whether each of these factors really does affect the degree of freedom in a particular act, which other dimensions should be added, when freedom on one dimension affects freedom on other dimensions, how to add these dimensions together to produce an overall judgment of freedom, what information we need to apply these dimensions to actual cases, and the implications of recognizing these degrees of freedom for our theories and methods, our personal lives, and our social policies and institutions. If you have time, we recommend that you re-read in advance the short chapters on “What is freedom?” and on “Does free will come in degrees?” in our joint book.
Controlling (volitional) decisions: towards a reductionistic framework to describe free will
Moderator: Walter Sinnott-Armstrong
In this talk I will start with a few wild and provocative conjectures, including with the notion that non-human animals have as much, or as little, free will as humans. To the extent that this is a valid conjecture, we can then use animal models to study the mechanisms of volitional decisions, and also to control volitional decisions. I will show preliminary evidence that it is possible to bias decisions in animals by directly tampering with their neural circuits. However, I will question the extent to which such control experiments illuminate the nature of free will. I will conclude by inviting all of us to brainstorm about what types of experimental evidence from control experiments could shed light on the neural circuits underlying volitional decisions.
Variants of preparatory signals and control: Interpreting computational models of decision processes
Moderator: Walter Sinnott-Armstrong
There has been a long history of investigating brain signals preceding voluntary action. Following the work of Benjamin Libet, many (including myself) have taken such choice-predictive signals as evidence that a decision can be made unconsciously by the brain before a person subjectively believes to be making the decision. Here, I will look into several conceptual issues regarding these early brain signals, and explain why I changed my mind. Two lines of work have challenged the interpretation of early signals, specifically of one variant, the readiness-potential. In experiments from our group we found that when readiness potentials arise in the brain, they do not necessarily lead to a movement. Instead, a prepared movement can still be aborted. This questions whether Libet-style arguments can really be used to question “free will”. In another line of work, Schurger and colleagues used a stochastic decision model that suggests that readiness potentials might be an artefact of reverse averaging of stochastic fluctuations in brain signals. I will report on our work on the model and our ongoing discussions. I will outline some misunderstandings regarding what this model does and what it does not claim, which aspects of the data are constrained by data, and how it models the different stages of decision making. Finally, I will discuss the conceptual question what it would take in such a highly dynamic system as the brain, to claim that a decision is made “early” or “late”.
Movement without preceding conscious intention of “what”
Moderator: Tim O’Connor
The goal of the consortium is to evaluate the role of consciousness in voluntary movement and decisions. Aim 2.5 compares right or left hand movements made in a choice reaction time paradigm where the triggering stimulus for what movement to make is masked by a congruent or incongruent stimulus. In the case of the incongruent mask, the subject sees only the mask, but might well respond to the initial unseen stimulus. Subjects believe their movements are voluntary and chosen prior to movement, so believe they made a mistake when they respond to the unseen stimulus. Behavior and EEG activity are monitored. Responses to the unseen stimulus in the incongruent condition are faster than those to the seen stimulus. Information flow in the brain shows that information flows rapidly from the visual cortex via the dorsal stream to the motor cortex where the response is initiated immediately. It appears that the information from the second stimulus arrives in the fronto-central region of the brain during the initiation of movement, but initiation is not halted. Instead, an error signal is generated that gives rise to the conscious sense that the movement was made opposite to the intention. It appears that, in this situation, movement choice is made with a sense of prior choosing, but that actual prior consciousness of choosing seems unlikely.
The role of consciousness in decision making and action formation: intentions, attention, the readiness potential, and the sense of agency (Part I)
Uri Maoz, Tomáš Dominik, Jake Gavenas & Alice Wong
Moderator: Tim O’Connor
The role of consciousness in decision making and action formation: intentions, attention, the readiness potential, and the sense of agency (Part II)
Uri Maoz, Lucas Jeay-Bizot, Martha Shaw & Jake Gavenas
Moderator: Tim O’Connor
Volition under uncertainty
Moderator: Pamela Hieronymi
Voluntary actions are distinct among human behaviours in several ways. They involve both an ordered series of neural events in the premotor areas of the brain, and also a characteristic conscious
experience of controlling one’s actions, i.e., the “sense of agency”.
Voluntary actions are normally guided by a combination of endogenous cues, e.g., external stimuli and
information, and exogenous factors, e.g., prior beliefs, desires, and “intentions”. Contrary to the
so-called “stimulus-driven actions”, the endogenous and endogenous factors associated with voluntary
actions are rarely deterministic. Instead, they can be characterized by a variable degree of uncertainty.
However, how the brain processes this uncertainty and combines exogenous and endogenous stimuli to
generate the appropriate voluntary action to accomplish the goal is still unknown.
In this project, a set of experiments explored whether the degree of endogenous and endogenous
uncertainty about the upcoming movement modulates the neural activity during motor preparation of
voluntary actions, as well as the behavioural experience of agency.
Does responsibility require consciousness?
Moderator: Pamela Hieronymi
Conscious perception entails phenomenologically experiencing external inputs in an accessible and reportable way. Although ubiquitous in shaping everyday experiences, there is still no psychological, philosophical, and neuroscientific consensus regarding its mechanistic underpinnings and functional significance in shaping human behavior. These debates are critical toward understanding the extent to which people need consciousness to evaluate, reason through, and apply morally relevant information to achieve responsibility-level responsiveness. In this talk, I will present research investigating (1) how the folk consider consciousness when evaluating an agent’s moral responsibility and (2) how the spontaneous neurocognitive state of the brain facilitates later conscious perception. I discuss this research toward adjudicating between two prominent theories of moral responsibility that do or do not consider consciousness as a necessary factor.
The role of conscious control in real-world decision making
Moderator: Pamela Hieronymi
Empirical studies of decision-making typically utilize paradigms where people are given a highly constrained set of options (usually two) to choose between. However, most of the actual choices we make, from mundane ones like what to have for lunch to profoundly important ones like the choice of a career, do not work like this. Instead, they require us to choose from poorly defined and often overwhelmingly large sets of options. Consider the seemingly limitless variations of meals one could have for dinner. In such cases, one cannot decide what to eat by first evaluating every option. Rather, one must first generate a constrained set of options to choose among. But how do we generate these “choice sets”? This is a critical aspect of real-world decision making that has been vastly understudied in both the free will and decision-making literatures. Prior research suggests that when generating choice sets, the options which come to mind are influenced by non-conscious processes, such as “model-free” estimates of how good that option has been in the past for similar decisions (Phillips, Morris & Cushman, 2019; Morris, Phillips, Huang & Cushman, 2021). While this work demonstrates that option generation is automatically influenced by sub-conscious processes (such as model-free reinforcement learning), we argue that conscious control must also play a central role in generating sets of options. In this study, we leverage the influence of value on the options that automatically come to mind to manipulate the role of conscious control in choice set generation. We hypothesized that when asked to make decisions involving selecting the best option, good options will automatically come to participants’ minds, and conscious control not be necessary for guiding choice set generation. However, when asked to make decisions involving selecting the worst option, conscious control will be necessary for guiding choice set construction. We find clear evidence that our paradigm successfully manipulates the role of conscious control in choice set generation: Participants take longer to make their decisions and rate these decisions as significantly harder when they are required to make decisions involving choosing low-value options. Our next step is to investigate the neural signature of this conscious control by having participants complete this task while in the scanner.
How can computational models help us understand free will?
Hans Liljenström, David Silverstein & Azadeh Hassannejad
Moderator: Aaron Schurger
Together with experiments and theory/philosophy, computational modeling is a central approach in our Neurophilosohy of Free Will project. In this conference session, we would like to discuss to what extent computational models actually can contribute to an understanding of free will, and which type of models are most likely to be helpful. In particular, we would like to present and discuss two current modeling efforts within the Agora research group:
1) Attractor neural network models of the neurodynamics of three cortical areas, LPFC, ACC, and pre-SMA, and their interaction. Here we focus on inhibitory and attentional control in self-initiated action, and investigate the impact of the interplay between different neural oscillators on the early RP. In addition, we study how coherent gamma frequency oscillations in the LPFC can trigger the initiation of an RP in pre-SMA, as part of a decision making process.
2) A computational model on possible subcortical contributions to the RP, which includes a dynamical systems representation of cortical-basal ganglia-thalamo-cortical loops during motor preparation. Preliminary results indicate that the early RP may be a preparatory loop while the late RP may be triggered by the release of response inhibition. This suggests that at least the early RP may not represent action in the absence of free will.
How do these models relate to other computational and experimental work and to the philosophical questions raised within the project, and how can we go forward in the remaining part of the project?
Is there a need for a meta-look on our field? Lessons from the ConTraSt database
Moderator: Tim Bayne
In this discussion, I will present the ConTraSt database for analysing and comparing empirical studies of consciousness theories. There, we analyzed 412 experiments that interpreted their findings in light of four leading theories of consciousness. As I will show, this analysis yielded several insights on the field at large and on potential biases in the way consciousness has been investigated. I will further present the open-access website we created, allowing users to conduct their own queries of the dataset. I will then ask if we, as a group, think that it would be beneficial to establish a similar database for the field of volition.
Uncovering the mechanism of intentional binding using visual masking
Aaron Schurger & Amber Hopkins
Moderator: Tim Bayne
If two events occur sequentially in time, a causal relationship may be perceived. If a causal relationship is perceived between two events, and the first event is a self-initiated movement, then the time interval between the first event (the cause) and the one that follows (the effect) is perceived to be shorter than it would be otherwise. This contraction of subjective time is known as “temporal binding”. The mechanisms responsible for this phenomenon remain unknown. Two explanations are possible: (1) temporal binding involves a change in time perception such that the time interval between the cause and effect is perceived to be shorter, even though the time interval between the brain events corresponding to the cause and effect is the same as it would have been otherwise; or (2) the brain events corresponding to the processing of the cause and effect actually do occur closer in time to each other, such that e.g. the neural response to the effect actually occurs earlier than it would have otherwise. We can refer to these two possibilities as an effect on (1) the perception of time or (2) the timing of perception. We set out to distinguish between these two possible explanations of temporal binding in terms of their respective effects on visual perception. We used visual backward masking to interfere with visual perception. Backward masking is the reduction or elimination of the visibility of a target (a brief stimulus such as a letter or shape) by presenting a mask (another brief stimulus) immediately afterward. The delay between the target and mask (called the stimulus onset asynchrony or SOA) is the main factor in determining the visibility of the target – shorter SOAs render the mask more effective (and hence render the target less visible). We tested the hypothesis that temporal binding increases the effectiveness of a visual mask when that mask is perceived as the effect in a cause-effect relationship.
Tillmann Vierkant & Jonathan Hall
Moderator: Manuel Vargas
A paradigmatic experience of freedom is the felt effort associated with the act of making a difficult choice.
In recent years, theories of mental effort have shifted from a focus on resource depletion towards a focus on cognitive resource allocation. Kurzban proposed that there is a cost-benefit computation, which compares the value of eg uncertainty reduction via continued deliberation to the value of using cognitive resources elsewhere. When the computation outputs the binary metacognitive decision to prolong deliberation, it also outputs an analogue feeling of effort which reflects the opportunity costs of continued deliberation.
We will argue that what triggers the making of a difficult choice is the agent’s (implicit or explicit) metacognitive decision to settle the question now, and act, rather than to prolong deliberation.
The natural symmetry of the metacognitive cost-benefit computation suggests that, rather than being limited to prolonging deliberation, Kurzban’s proposal can be extended to effortfully stopping deliberation. Here the effort felt is a reflection of the opportunity cost of settling the question now, despite ongoing uncertainty.
This raises an interesting question: Could neuroscience provide an objective marker of subjective mental effort in the decision-making process, perhaps by evidencing shifts in baseline-to-threshold distance and/or gain-modulation?
Conscious volition as a test case for theories of consciousness
Tim Bayne & Liad Mudrik
Moderator: Manuel Vargas
The field of consciousness studies has recently put special emphasis on comparing and testing theories of consciousness. Yet these tests have so far solely focused on conscious perception; that is, the processes that unfold when an external stimulus is consciously (vs. unconsciously) perceived. Here, we claim that theories of consciousness should also account for other types of conscious experience, and suggest that conscious volition might serve as an interesting test for these theories. We begin by clarifying conscious volition, focusing on how it differs from other aspects of consciousness, such as perceptual experience. We then examine key theories of consciousness, asking how – if at all – they might account for key features of conscious volition. Asking how theories of consciousness might be extended beyond perception and applied to conscious volition is crucial not just to a comprehensive understanding of consciousness, but also to a full understanding of voluntary agency.
Unconscious reactions and reactions to unconscious stimuli
Tomáš Dominik, PhD
We spend much of our lives monitoring our environment and reacting to stimuli when appropriate. In our experience, we are conscious of both the stimulus triggering our action (stimulus awareness) and the action itself (action awareness) in these situations. If, as you are sitting down and reading this text, you would find yourself standing up without knowing why and/or how that came to be, it would likely be truly disturbing. It might thus seem that stimulus awareness and action awareness necessarily accompany (or at least follow) any stimulus-reaction event. But is this always the case, universally true for any stimulus and any reaction? Our project aims to shed light on the divide between behaviors that can and cannot manifest without these types of awareness. We hypothesize that at least certain simple reactions can take place with neither stimulus awareness nor action awareness. In our experiment, we used a go/no-go paradigm with both supraliminal and subliminal stimuli. We found that supraliminal no-go stimuli sometimes invoked a measurable muscle activation—clearly in response to the stimulus—but with the subjects reporting being unaware of any response whatsoever. We also found that subliminal go stimuli significantly increased the muscle activity even when the subjects reported not being aware of any stimulus. Our results therefore suggest that neither action awareness nor stimulus awareness is necessary for stimulus-driven action initiation
Perturbing the Sense of Agency with closed-loop TMS-EMG
Much of the neuroscience of volitional action literature investigating temporal aspects of action awareness compare the time courses and neural correlates of externally triggered action to internally, self-generated action. Existing studies have also looked at the effect of providing different combinations of delayed, false, visual, and auditory feedback on the movement and awareness of movement times. However, to the best of our knowledge, investigations on the quality of awareness of action and sense of agency over actions where the intention to act is self-generated, yet a bodily movement is simultaneously externally triggered, has been lacking. How might the sense of agency be perceived under such circumstances?
We developed a closed-loop EMG-TMS experiment to investigate this question. First, we found a hand movement that could be reliably triggered in the participant by TMS to the contralateral primary motor cortex. We then instructed the participant to make that movement at a time of their choice. We then used the EMG activity that led up to the finger movement to trigger TMS that elicits this specific movement. Interestingly, when asked, subjects reported that they did not initiate the finger movement, instead reporting that the computer initiated the movement, or that both the computer and they themself initiated the movement.
What neural network features and dynamics underlie self-initiated action?
Stochastic accumulator models of self-initiated action posit that slowly-ramping signals that emerge before movement (such as the cortical readiness potential, RP) correspond to a diffusion-to-threshold process, where the threshold-crossing is related to movement timing (e.g. Schurger et al., 2012). Pre-movement ramping is also seen in the firing rates of individual neurons in the SMA (Fried et al., 2011), but autocorrelation timescales of individual neurons are too low for a single-neuron diffusion process to account for this ramping (Cavanagh et al., 2016), implying that if stochastic accumulation occurs, it occurs at the network level. However, what network features could give rise to these gradual changes in firing rates is unclear. By simulating networks of spiking neurons, we demonstrate that clustered architectures combined with slow synaptic dynamics are sufficient for a network to show pre-threshold ramping of single-neuron firing rates as well as a gradual negative shift in an EEG proxy signal, reminiscent of the RP. Furthermore, by including both slow and fast synapses in the network the times of threshold crossings closely matches those found experimentally. Finally, we show that gradual ramping is only one of at least three types of pre-movement changes in network activity through a low-dimensional single-trail analysis of intracranial data.
Baseline correction may have been responsible for false results in decades of readiness potential research
Among the known antecedents of spontaneous voluntary actions, the readiness potential (RP), a slow negative deflection in the electroencephalographic (EEG) over the motor cortex, is the most widely studied signal. The RP extends over hundreds of milliseconds up to multiple seconds before the onset of a spontaneous voluntary movement. Where to place the onset of the RP remains open to debate. Recent interpretations argue that the RP’s onset reflects a decision process that originates as early as trial onset, which can be several seconds prior to the movement. When processing EEG data, it is standard practice to subtract a baseline from the signal as a data preprocessing step, in order to correct for slow drifts and offsets in the signal. In RP research, it is standard practice to subtract the mean amplitude found in the first 100ms of the epoch as the baseline. This places the baseline period anywhere from a few hundreds of milliseconds to multiple seconds before movement. Baseline correction relies on the assumption that the activity of the baseline period is independent from the task. Using a statistical argument we demonstrate that the primary assumption of baseline correction, independence of the baseline period from the signal, is violated in RP research. Given that the RP may start in this early time frame, baseline correction could, arbitrarily, determine the results. Using simulations and a meta-analysis, we further show that, when comparing the RP across conditions, baseline correction can generate false positives and even invert the interpretation of the results. Finally, we consider and propose alternative processing steps and advocate for a shift away from time-locked analysis in volition research.
Comparison of Causal Modeling Methods on MEG and fMRI Data of the Human Connectome Project
Dehua (Andy) Liang
Neuroscience research is often concerned with understanding the brain regions and their interactions associated with a certain cognitive process of interest. These interactions are often studied using functional connectivity, a technique that investigates the neural time-series correlation and coherence. Recently, Reid et al. called for a shift of functional connectivity research from association to causation (Reid et al., 2019). Recent development of multivariate analysis methods (Haufe et al., 2014) also stresses the importance of interpretability and the underlying data generating processes of neural signals. Dubois et al. proposed a causal modeling framework to discover causal brain connectivity for emotions (Dubois et al., 2017) using fMRI data from the Human Connectome Project (HCP) (Van Essen et al., 2013). While the method showed promising results by obtaining causal graphs on high-spatial-resolution fMRI data, its low temporal resolution prevents the analysis of neuronal oscillations and coherence (Hall et al., 2014). In addition to fMRI data, the HCP dataset also provides MEG data for 95 subjects along with head and source models. We are developing a processing pipeline to discover the causal connectivity of that MEG data, comparing with and validated by the fMRI data. The effect of temporal and spatial information in causal discovery algorithms is evaluated using this empirical approach.
Free Will: easy or hard problem?
Alessandra Buccella, PhD
In her article How does neuroscience affect our conception of volition? Adina Roskies writes: “The mere fact that […] volition has neural underpinnings is […] not a basis for denying freedom of the will” (Roskies, 2010, 123). How to interpret this claim? In this poster, I offer a tentative argument for what I call the “Hard problem” interpretation. Roughly, the hard problem interpretation of Roskies’s claim is one according to which, even if the neuroscience of volition led us to re-think the nature of the homonymous faculty, it won’t be able to make us re-think our conception of Free will, since those actions for which the question about freedom makes sense are not the actions studied by the neuroscience of volition. My argument relies on Uri Maoz and colleagues’ experimental evidence that the “Readiness Potential” (RP), i.e. arguably the most important and widely discussed piece of evidence within the neuroscience of volition, is not observed in “meaningful” decisions (Maoz et al., 2019) and on G.E.M. Anscombe’s view of intentional actions and reasons for acting (Anscombe, 2000). According to Anscombe, an answer to the question “why?” given with respect to an action is a reason for acting if “in treating it as a reason [the agent] conceives it as something good or bad, and his own action as doing good or harm” (Anscombe, 2000, 22). Maoz et al. (2019) have argued that actions involving deliberation and the assignment of values to possible outcomes, are not preceded by an observable RP in the EEG signal, which in turn suggests that they are not rooted in the same mechanism as arbitrary decisions normally used in Libet-style experiments (cfr. LIBET et al., 1983). If, as I find plausible, actions relevant for the Free will debate are done for reasons (in Anscombe’s sense), they are also not analyzable in terms of the volitional mechanism centered on the RP.
Freedom from what? A componential analysis of volitional behaviour
Keiji Ota, PhD
Volition is the cognitive process by which agents endogenously initiate actions without being triggered by external stimulus. One of the key features of volition is spontaneity which refers to a capacity for generating innovative actions. Such voluntary control induces variability in behavioural choices while preventing habitual responses. This capacity becomes critical when agents explore a reward landscape in the environment, and when they need to avoid predators. However, investigating this aspect of volition is challenging because human subjects show poor performance in responding randomly even if they are explicitly asked to do so. One possibility is to expose subjects to different situations where they must randomise behavioural choices in order to get reward. Methods. Here we sought to develop a new experimental paradigm by confronting subjects with a virtual competitor that attempts to predict their future choice, and then rewarding them only when they avoid predictability. In a laboratory experiment, twenty participants viewed birds resting on a tree. In each trial, the participants were asked to decide the time they threw food. If the time participants threw matched the time the birds flew out of the tree, the food was eaten, and the participants lost. If the participant’s action time did not match the birds’ flight time, the participants avoid getting their food eaten by the birds, and thereby win. Importantly, the competitor (birds) could learn when the participants would act based on past behaviour. Thus, the participants had to produce the variability in the timing of their actions to avoid the competitor’s prediction. Results. Three separate virtual competitors were programmed so that they had increasingly sophisticated predictive power of reading four habitual responses. 1. Choice bias: the competitor could predict the tendency to respond early or late. 2. Transition bias: any choice patterns in the relation between one action to the next were predicted. 3. Reinforcement bias: any choice patterns in the relation between one outcome to the next action were predicted. For participants to perform successfully again each class of competitor, they need to act in a way that is even more unconstrained than required by the preceding class of competitor. Thus, participants require different elements of ‘free’ action in a structured series. We measured the statistical distance between the probability distribution of the participant’s choice pattern and the probability distribution of the ideal observer’s pattern. We found that the statistical distance after loss trials increased compared to the distance after win trials. Discussion. The results showed that the negative reinforcement was the primarily component which limits the freedom of choices. People were more successful in avoiding any regular transitions following the positive outcome. However, they failed to become less dependent on the negative outcome. These results suggest that our paradigm can effectively examine the spontaneous aspect of volition in a self-paced movement, and that a cognitive hierarchy of different forms of freedom underlies the innovation aspect of voluntariness. This paradigm allows systematic investigation of the different cognitive components of endogenous action initiation.
On-line and inferential contributions to the subjective time of decision
Bartosz Majchrowicz, PhD
Research on human volition has shown that the experience of intending or deciding to do something is influenced by both prospective and retrospective/inferential processes. However, few experimental studies have successfully dissociated on-line readout of internal (mental) states from action- and feedback-related inferences. Therefore, the contribution of each process to the subjective experience of decision, and to the electrophysiological markers of decision (CPP, RP, P3) remains unclear. We report an experimental study that assesses these contributions, and relates them to decision models that assume separate evidence thresholds for decision and for motor responding.
This study uses evidence-based decision making task combined with mental chronometry. In this task, participants monitor external evidence under two difficulty conditions and have to report which evidence is dominant. Report is done by either pressing a key or withholding to do so. Feedback presence is also manipulated. Thus, factorial manipulation of evidence, action, and feedback is employed. Concurrent EEG recordings are analysed to extract classic event-related potentials components linked to decision-making.
Study is currently in the data analysis stage. Its behavioural and electrophysiological results will be presented for the first time during the meeting. Results will be discussed in relation to the current evidence accumulation models, as well as optimal cue integration models. Assessing the relative contribution of external evidence, action presence, and feedback presence to the subjective decision times will have important implications for current research on volition and decision-making, by shedding a new light on factors modulating subjective experience accompanying free choices.
Voluntary action and problem solving
Silvia Seghezzi, PhD
Volition refers to the mental capacity to initiate actions through one’s autonomous decisions. Scholars disagree over definitions of volition. Further, existing experimental paradigms for studying volitional action focus on arbitrary, meaningless actions and on their neural precursors. In contrast, this project develops a new view of volition as means-ends problem solving. We study how volition contributes to the organisation and execution of complex goaldirected action sequences, thus linking volition to classical neuropsychology of executive functions and problemsolving. The project considers volition as a ‘smart’ cognitive function, providing an alternative to current views of volition as either randomness, or subjective preference in evaluative decision- making. Here, a structured series of three experiments used “Tower of London” tasks to identify distinct cognitive processes related to volition, such as planning, choice, and inhibitory control.
A protocol to investigate the brain correlates of responses to unconscious stimuli
Antonio Ivano Triggiani, PhD
To investigate the role of consciousness in voluntary movement, it would be useful to have a paradigm to produce movements, believed to be voluntary, in response to stimuli that are not seen. A common way to modulate awareness in neuroscientific protocols is the masking effect to hide the presence of a stimulus on a screen. It has been demonstrated that these stimuli can be processed unconsciously, inducing a change in behavior such as making a movement. If it is a choice reaction time paradigm, it is necessary to identify the stimulus in order to make the right movement. In a simple reaction time paradigm, it would not be necessary – the awareness of something happening would be enough to trigger a prepared response. The electro- (EEG) and the magneto- (MEG) encephalography are the best devices to analyze the brain correlates to those stimuli in the necessary ms range. These kinds of tasks present several issues. One of the main problems is the presence of different stimuli on the screen leading to complex potentials that need to be untangled. Another issue is the effectiveness of the masking effect and its variability across the participants.
In this study, using some pilot trials, we tried to build a protocol to analyze the behavior and the EEG brain correlates of masked stimuli. We used a choice reaction time for moving the right or left hand, and sometimes the mask was opposite to the initial stimulus. We tested the protocol, with a final experiment enrolling 20 participants (25-41y) undergoing EEG recording during three different parts. The three tasks were: the main experiment, using backward masking with metacontrast; a “mistake” questionnaire aimed at testing the perception of the behavioral response; a test to measure the Perceptual Awareness Scale (PAS). The main experiment showed that subjects did make some responses to the masked stimulus. The mistake questionnaire confirmed that while subjects responded to the initial stimulus, they actually saw the masking stimulus and therefore thought their voluntary movement was an error. The PAS confirmed that often subjects did not see the initial stimulus at all, but sometimes did see something, but did not know what it was. These data taken together confirm that we were able to produce movements to stimuli, processed by the brain, but not perceived. Analysis of the EEG would then let us see what happens in the brain in this circumstance and what happens in the brain when the response is in accord with the masking stimulus and likely perceived.
Identifying the underpinnings of volitional control through event-related microstate analyses of seen and unseen stimuli
The role of consciousness in volitional control has yet to be elucidated. The analysis of microstates, which are brief quasi-stable brain states, may be a useful technique for investigating the brain states that underlie conscious and unconscious processes. It has been postulated that there are correlations between microstates and qualia, the subjective elements of consciousness. If so, such an analysis will lead to a greater understanding of the events underlying consciousness during movements and decisions. This topic was investigated by manipulating whether stimuli were consciously perceived or not perceived. Through a choice reaction time task, participants were presented with two sequentially presented arrows and were instructed to respond to their directionality. In the congruent condition, the directionality of the arrows was the same, in the incongruent, the opposite. The first arrow was backward masked by the second one, using the metacontrast technique. In the incongruent case, the participant only saw the second stimulus but could make a voluntary motor response to the unseen first stimulus. Responses to both seen and unseen stimuli in the incongruent condition and to the stimuli in the congruent condition gave rise to an initial microstate in which the occipital cortex received information about the presented stimuli; this microstate contains the event-related potential (ERP) component N100. This was followed by a microstate with negative lateral occipital distributions related to processing in visual association areas; it contains the visual awareness negativity (VAN) and may initiate visual awareness. The second microstate also includes some frontal positivity and may therefore represent a combined state of the quale of visual awareness and movement initiation driven by vision. The third microstate is typically the same as the first microstate and may well indicate further visual processing and/or awareness. The fourth microstate is longer than the others, occurs well after the movement, and may include an evaluation of what just happened. It is interesting that the pattern of microstates is similar in all our conditions, which presumably indicates similar brain processes. It is still not clear whether we can use this analysis to precisely time qualia.
Suppress me if you can: Neurofeedback of the readiness potential
Matthias Schultze-Kraft, PhD
Voluntary movements are usually preceded by a slow, negative-going brain signal over motor areas, the so-called readiness potential (RP). To date, the exact nature and causal role of the RP in movement preparation have remained heavily debated. Although the RP is influenced by several motorical and cognitive factors, it has remained unclear whether people can learn to exert mental control over their RP, for example, by deliberately suppressing it. If people were able to initiate spontaneous movements without eliciting an RP, this would challenge the idea that the RP is a necessary stage of the causal chain leading up to a voluntary movement. We tested the ability of participants to control the magnitude of their RP in a neurofeedback experiment. Participants performed self-initiated movements, and after every movement, they were provided with immediate feedback about the magnitude of their RP. They were asked to find a strategy to perform voluntary movements such that the RPs were as small as possible. We found no evidence that participants were able to to willfully modulate or suppress their RPs while still eliciting voluntary movements. This suggests that the RP might be an involuntary component of voluntary action over which people cannot exert conscious control.
The geometry of domain-general performance monitoring in the human medial frontal cortex
Zhongzheng (Brooks) Fu, PhD
Controlling behavior to flexibly achieve desired goals depends on the ability to monitor one’s own performance. It is unknown how performance monitoring can be both flexible to support different tasks and specialized to perform well on each. We recorded single neurons in the human medial frontal cortex while subjects performed two tasks that involve three types of cognitive conflict. Neurons encoding predicted conflict, conflict, and error in one or both tasks were intermixed, forming a representational geometry that simultaneously allowed task specialization and generalization. Neurons encoding conflict retrospectively served to update internal estimates of control demand. Population representations of conflict were compositional. These findings reveal how representations of evaluative signals can be both abstract and task-specific and suggest a neuronal mechanism for estimating control demand.
What will the neighbors think?
– Discussing volition with experts from neighboring fields
Ralph Adolphs, Robyn Waller, Bill Newsome & Shin Shimojo
Moderator: Adina Roskies
Research on Free Will/Consciousness at NIH
Moderator: Uri Maoz
Dr. Gnadt will survey the active research awards at NIH on consciousness and related topics, and will discuss opportunities at NIH for funding neurobiological, social/behavioral and ethical aspects of consciousness as a cognitive process.
TWCF and Team Science: Adversarial Collaboration, Open Science and Beyond
Moderator: Uri Maoz
TWCF has a very keen interest in supporting team science. We have been developing unique and innovative grant making mechanisms that encourage collaboration between groups including our Structured Adversarial Collaboration in our Accelerating Research on Consciousness Initiative. We are also promoting best practices in open science across all of our grant development and continuously exploring various approaches to encourage collaboration.
|Chapman||Dehua (Andy) Liang|
|Indiana University Bloomington||Tim O’Connor|
|Charité – Universitätsmedizin Berlin||John-Dylan Haynes|
|Charité – Universitätsmedizin Berlin||Matthias Schultze-Kraft|
|Boston Children’s Hospital||Gabriel Kreiman|
|Sigtuna Foundation||Hans Liljenström|
|Sigtuna Foundation||Alf Linderman|
|Sigtuna Foundation||David Silverstein|
|Tel Aviv||Liad Mudrik|
|Tel Aviv||Yarden Shir|
|Pomona College||Shlomi Sher|
|Templeton World Charity Foundation||Virginia Cooper|
|Iona College||Robyn Waller|
|Université Libre de Bruxelles||Guillaume Pech|
|Université Libre de Bruxelles||Emilie Caspar|
|Kaunas University of Technology||Paulius Rimkevičius|
Need a COVID test before your flight back?
Congratulations to conference attendees Alice Wong and Cara Lewis, who were featured as Chen Science Writers by the Tianqiao & Chrissy Chen Institute. Their meeting reports were published on the Tianqiao&Chrissy Chen Institute Newsroom.
Organizers: Uri Maoz & Tian Lan