In this paper, I have discussed some scientific research on consciousness and conscious (free) will, focusing on a series of neurophysical experiments, which have been used in the debate as evidence supporting free will as an illusion. In particular, I have discussed temporal aspects of willed actions and referred to experiments that attempt to associate the timing of neural events with such actions, and where subjects are able to report about their subjective experiences. Neurophysiological techniques, such as EEG are especially suited for this, as it has a high temporal resolution and is fairly easy to apply. These experiments are largely based on the findings that some “signal”, e.g., the so-called readiness potential (RP), precedes a conscious will to perform a movement.
I have further referred to brain imaging experiments, such as PET and fMRI, which are based on the relation between cortical blood flow and neuronal activity. These methods have a high spatial resolution, but a poor temporal resolution, and with the drawback of reflecting inhibitory activity, indistinguishable from excitatory activity. Brain imaging has revealed that a number of brain regions contribute to the performance of willed actions, in particular prefrontal cortex together with those brain regions with which it is connected (Spence and Frith 1999). Disease or dysfunction of these circuits may be associated with various disorders of volition, including “alien control”. I have also briefly described some experiments, where subjects are responding to electric or magnetic stimulation applied directly to the brain.
When carefully examining the experimental procedures and results discussed above, I can find no convincing evidence supporting the illusory conscious will (ICW) hypothesis. Similar conclusions have been made by, e.g., Heisenberg (2009), Sternberg (2010), Mele (2009, 2019) and Lindahl and Århem (2019). Further, recent studies by Schurger et al. (2012), Maoz et al. (2019) and Mudrik et al. (2020) even seem to undermine arguments for ICW based on previous interpretations of the Libet experiments. Batthyany (2009) not only questions the ICW hypothesis, but also points at the general bias in the interpretation of experiments, where the dominating philosophical preference abolishes alternative interpretations.
The alternative hypothesis of causative conscious will (CCW) is also neither falsified nor confirmed by the evidence. In fact, it is not an easy task to design and perform experiments that could reveal the true nature of willful acts, especially not in an artificial environment with non-ecological tasks. One could argue that experiments like those of Libet and Soon et al. do not test for free will at all, since the subjects are only asked to perform an “action” when there is an urge to move, and these movements can be said to be actions only in a very limited sense. Indeed, the way we pose our questions, set up our experiments, and instruct our subjects, is guided by our preconceived beliefs and assumptions, and hence it is difficult to get results that would contradict a dominating paradigm.
There is a great need for refined experiments and an unbiased analysis of the empirical evidence, which better can address the problem of conscious (free) will in natural complex situations. In particular since beliefs about our conscious will can have a significant impact on people’s moral outlook and behavior (Vohs and Schooler 2008; Sternberg 2010). The field also urgently needs a more precise and consistent terminology that avoids ambiguity and minimizes confusion (regarding concepts such as will, volition, intention, decision, and choice), although some attempts have already been made (e.g., Mele 2009). Such a terminology, if consensus about it could be reached, would facilitate interpretation and communication of hypotheses and experimental results.
In a newly started interdisciplinary project, The Neurophilosophy of Free Will (www.neurophil-freewill.org), we are actually attempting to sort out the various concepts related to consciousness and free will, as used in the literature. We also try to construct and carry out more ecological, realistic experiments, where initial results indicate that no RP appears for deliberate decisions in voluntary actions, as it does for arbitrary decisions (Maoz et al. 2019; Mudrik et al. 2020). In addition, we are using computational models that aim to suggest neural correlates and causative pathways included in decision making and volition.
In this paper, I have briefly described our initial computational modeling of decision making (DM), as an important part of volition. We have modeled various brain structures involved in DM, including the amygdala, the orbitofrontal cortex and the lateral prefrontal cortex, representing emotional, as well as cognitive aspects of the DM process. With our model we have demonstrated how different cell assemblies, representing different optional choices available to the individual may compete with respect to the level of activity. This level was suggested to be a combined measure of the size of the cell assembly (number of neural populations), and the frequency and amplitude of the oscillatory activity of the neural populations. The different options get different values, depending on internal as well as external factors. The “winning” assembly was simply the one with the largest neural activity, resulting in a decision, a choice for that option (with some probability).
A good model can help in understanding a system or process, predict the outcome of an experiment, or guide in new experimental studies. A model is, however, never correct, it is only more or less useful depending on the purpose. As long as the model output is interpretable in experimental results, or in a theoretical hypothesis, it may serve its purpose. Yet, when constructing a computational model of any neural system or process, we have to make tremendous simplifications. The problem is to find an adequate simplification, where the essential details are included, and the less relevant ones neglected. Indeed, one of the greatest challenges in the modeling process is to extract the relevant details out of the enormous amount of known facts about brain structures and their dynamics. Just simulating a neural process, or a cognitive function can never give any conclusive understanding of the system or process being modeled, just suggesting which solutions seem most plausible.
In our computational model, as in reality, there are many internal and external factors that all the time influence the DM process. In the complex neuronal networks of the brain, with many parts that interact and influence each other in myriads of feedforward and feedback loops, it is very difficult to determine any causal pathways, or what can be considered the first “independent” signal initiating the process leading to a decision, and subsequently to an action. Yet, neurocomputational models, such as ours, can serve as helpful tools when trying to describe and understand the underlying neural processes involved in DM, as a central part of volition and the exercise of free will. (In our further modeling, we also try to include effects of social interactions, which often are neglected in studies related to free will).
Decisions are based on an integrated evaluation of different emotional and cognitive assessment of the consequences of the decisions/actions, but it is not certain that a decision follows automatically the valuation made by the system. In our model, a decision is taken only with a certain probability, which is given by a random generator. In reality, it could be our more or less free will which allows for a possibility that we (as conscious subjects, see below) do not have to slavishly follow what the brain has calculated as the best decision in each case, which in general could be considered the most rational.
Undoubtedly, our decisions depend on, but are not determined by biological (genetic, neural, physiological) factors. They also depend on psychological, social and environmental factors, which altogether constitute a complex web of causation. This makes it hard or even impossible to predict an action or behaviour for any given individual, when studied “from the outside.” The apparent unpredictable outcome could be interpreted as a result of random neural activity, exemplifying upward causation, or as a result of free will, if downward causation is considered.
While there are no convincing scientific arguments for either ICW or CCW, based on experiemental evidence, it appears that ICW is easier to defend from a scientific point of view, because it seems to fit with the dominating paradigm. Consciousness may there be considered an emergent phenomenon, but with no causative power on the underlying neural processes, which would require downward causation. In traditional, reductionist science, such downward causation is difficult to accept. Moreover, neither randomness nor deterministic laws of nature, which are the only scientific explanations available for what happens in the world, seem to allow for free will, as mentioned in the Introduction.
On the other hand, CCW seems to fit better with our experience and our social/cultural traditions, where responsibility for our actions appears to require a capacity to consciously choose between alternative actions. So, is there a way out of this dilemma? Could there be any room for free will in current science, after all? I concur that downward causation is necessary but not sufficient for demonstrating the existence of free will. However, in order to demonstrate downward causation, one has to show that a change in some high level variable(s) reliably results in a change in lower-level variable(s).
For a nervous system, as for complex systems in general, different phenomena appear at different levels of aggregation. Emergent phenomena may result from an upward, “bottom-up” causation, based on micro level phenomena. Yet, higher macro levels may also “control” lower ones (c.f. the so-called enslaving principle of Haken 1983), an example of downward, “top-down” causation. This interplay between micro and macro levels is part of what frames the dynamics of neural systems, and is another example of circular causality, which Freeman referred to for the action-perception cycle (Freeman 1999; see also the section on intentionality above). Of special interest is the meso level, i.e., the level in between the micro and the macro, where bottom-up meets top-down.
Mesoscopic brain dynamics is partly a result of a dynamic balance between opposing processes such as inhibition and excitation, which often results in oscillatory and chaotic-like behaviour (Freeman 2000; Liljenström 2012). Yet, this dynamics is mixed with noise, generated at a microscopic level by spontaneous neural activity. It is also affected by macroscopic activity, such as slow rhythms generated by cortico-thalamic circuits or neuromodulation from different brain regions. Effects of arousal, attention, or mood, through neuromodulation or other means, could be seen as a top-down interaction from macroscopic activity to mesoscopic neurodynamics. Our computational models have demonstrated how complex neurodynamics of cortical networks can influence the neural activity of single or populations of neurons. We have also shown how neuromodulation and attention can synchronize and in other ways regulate the activity of neurons and neural populations in networks (Liljenström 2016).
While the behavior of single molecules or cells may appear stochastic and noisy at micro- and mesoscopic levels, their collective behavior could generally be regarded as deterministic and ordered at the macroscopic level, where the irregularities at lower levels are “averaged out”. Still, under certain circumstances, single events at these lower levels may be amplified and cause state transitions or other phenomena at macroscopic levels. For example, retinal absorption of single photons under extremely dark conditions, or odor receptor absorptions of single odorous molecules can be amplified by various neural networks to result in conscious perception.
In addition, the complex neurodynamics of cortical networks may result in unpredictable chaotic behavior with a high sensitivity to “initial conditions”, so very small differences in cortical firing patterns may change the activity at cortical network levels and result in completely different output signals. Both experimental studies (Freeman 2000) and computer simulations (Liljenström 1995) indicate that the complex neurodynamics of cortical systems can provide chaotic intentional states, which eventually may converge to meaningful percepts. Mental processes may arise from neural activity, but they may also affect the neural activity, in a kind of downward causation that seems necessary for free will. The intricate web of inter-relationships between processes at different organizational levels of neural systems seems able to provide both upward and downward causation.
All of this might provide possibilities for several optional outcomes, “choices” of any brain-state. It may constitute necessary, but not sufficient conditions for free will. There also needs to be “someone”, a subject, to make the choices, in order for free will to exist. The subject may well be formed by the global activity of the brain-mind system, under constraints given by the physical structures, but it should be in control of its actions.
Indeed, the main reason why science has problems encompassing free will is that it seems to require the action of a conscious agent, and (natural) science has so far only dealt with objects, not agents/subjects. The theories and laws of physics were developed for inanimate, comparatively simple objects and their interaction, and have very little to say about the behavior of complex biological systems, in particular regarding brain-mind systems and their (inter-)actions. In fact, Einstein, as well as Schrödinger (1944) recognized the insufficiency of contemporary physics to describe the immense complexity of living systems. In lack of conclusive experimental evidence, perhaps theoretical insights, such as Einstein’s of space and time as non-separate and interdependent, could consider mind and matter as non-separable, and consciousness and agency as an essential part of our natural world. Indeed, when intentional or conscious actions include also the choice of mating partners, the caring for offsprings, and relational behavior in general, it is clear that consciousness may not only have effects at an individual level, but can also be seen as a driving force in evolution.
To conclude, we are embedded in this world in an ongoing action-perception cycle, continuously interacting with our environment. Causal relationships are difficult to determine, and external and internal influences (preferences, expectations, intentions, emotions) affect our unconscious and conscious actions. This is particularly obvious in a social context, in our interaction with others. However, if we should be accounted responsible for our actions, it appears obvious that we must be able to choose between different actions, that we must have a free will. Until we have more evidence to make any scientific conclusions about the existence of free will, it may be wise to be humble and rather rely on our intuition than on any counter-intuitive hypothesis that seems reasonable, just because it fits with the current paradigm. In order to allow for consciousness and free will, science probably needs to be extended beyond chance and necessity, which currently are its only models of explanation.