Overview of some existing demarcation criteria
In recent years, many philosophers and neuroscientists have explored demarcation criteria that could be used in deciding whether a being or, locally, a specific mental state is conscious. Here is a tentative list of common proposals:
(1) Subjective or introspective criteria refer to first-person verbal reports. Introspective criteria are traditionally considered sufficient to decide whether a being or a mental state is conscious. However, it is also accepted that subjective verbal reports present major epistemological limitations, as they are often erroneous (Carruthers, 2011), non-verifiable, and restricted to verbal human beings.
(2) Behavioral or agentive criteria assume that if the behavior of a creature is intentionally controlled, then consciousness is present, for example, that the fleeing mouse is aware that the cat is approaching, that the monkey pressing a button is aware of the objection in its visual field, or that the child placing a wager in "post-decision wagering" (PDW) experiments is aware of his experience. By using a broad concept of agency which includes mental agency (e.g., D. Owen's tennis-imagery studies on vegetative patients), one could even consider agentivity a valid criterion for all candidates to consciousness (incl. sedated/comatose patients, locked-in patient, AI, and brain organisms). It is unclear, however, whether agency is a sufficient criterion for consciousness: It has been shown that the behavior of a creature can be dissociated from its conscious experience, for instance, in epileptic seizure cases, or in “agentive fragmentations experiments,” where participants' behavior seems to contradict their states of consciousness (see, e.g., Cummings' studies in which participants, who were asked to press a button when they saw a specific letter on a screen, reported being aware of having made mistakes despite their behavior being correct). Other examples include split-brain cases, where patients deny seeing stimuli presented in their left visual fields while nevertheless spontaneously and rationally using representations of the objects presented in their left visual field (Shea & Bayne, 2010).
(3) Physiological criteria draw on neural and cognitive theories of consciousness. They have been inventive in identifying new criteria that may determine where and how a mental state is conscious. For example, the activation of a large global network in the human brain (GWT), attention to intermediate-level representations manifested by the firing of vectorwaves in the gamma range (AIR), maximum of irreducible integrated information measured by phi (IIT) have been proposed as neural criteria for consciousness. Some can even be considered promising for investigating cases where no compliance can be expected from an individual (e.g., hemispherotomy). For instance, the perturbational complex index (PCI), a metric of a technique involving transcranial magnetic stimulation to perturb the cortex, has been suggested as a consciousness indicator (Casarotto, 2016). It remains unclear, however, whether these physiological markers are sufficient or necessary criteria. Against GWT, for example, some studies have proven that non-global workspaces, say, fractured workspaces in epilepsy cases, can be considered conscious, suggesting that global availability isn't necessary for consciousness (Bayne, 2011). Other research suggests that global availability may not be sufficient either (Soto & Silvanto, 2016). Moreover, some physiological criteria can be considered contradictory. For instance, contrary to GWT, AIR restricts conscious experiences to attended intermediate-level representations (i.e., perceptions) and therefore exclude that high-level representations (e.g., abstract thoughts) can be conscious. Similarly GWT and IIT disagree on the cortical areas that are mainly responsible for consciousness: frontal and prefrontal area for GWT, posterior cortical areas for IIT etc.).
(4) Philosophical theories of consciousness, in particular representationalist theories, have been active in developing many theoretical criteria for detecting consciousness in recent years: intentionality (representationalism in general), first-order perception (e.g., Tye, Dretske), higher-order perception (e.g., Armstrong, Lycan), higher-order thought (Rosenthal, Carruthers), or self-representation (Kriegel). Although very stimulating, these criteria , which don't formally require empirical corroboration, can be considered speculative. It should be noted, however, that representationalists also pay attention to introspective, behavioral or physiological corroborations.
(5) The unlimited associative learning (UAL) criterion rests on the idea that conscious beings perform a type of learning with the following features: (1) learning from compound stimuli and novel stimuli, (2) second-order learning, (3) unlimited learning (involving more learning possibilities than could be had in one's lifetime), (4) trace conditioning (i.e., learning involving a time gap between conditioned/unconditioned stimulus), and (5) easy reassignment of the perceived value of a given stimulus if that stimulus becomes harmful (Birch et al, 2020). Proponents of the UAL criterion hold that while the spinal cord (if it's disconnected from the brain) can perform some simpler types of learning, it doesn't display UAL, which is taken to be sufficient for sentience (incl. phenomenal sensory experience, vision, olfaction, pain, hunger, etc.) and has important implications for distribution of consciousness in the animal kingdom. However, questions remain as to whether consciousness is actually required fo complex forms of leaning, and, for instance, higher-order thought theorists hold that the UAL criterion is too lenient (Mallet, 2021).
(6) Finally, many searchers today advocate a pluralist approach. Rather than restricting themselves to the use of a single type of criteria, or even introducing a hierarchy between the criteria, they claim that one must be contextualist and test, on a case-by-case basis, different criteria (e.g., Shea, 2012; Shevlin, 2019).
In short, different candidate demarcation criteria exist. Some of them are potentially conflicting with others. The objective of the “Towards new demarcation criteria for borderline consciousness” project is to critically analyze the existing proposals, identify their shortcomings, and develop new criteria that are able to overcome these limitations.