The Top Ten Criteria for Defining Consciousness

Ordered by consensus. A research report companion to Compassion and Consciousness.

Across philosophy of mind and contemplative traditions, the criterion with the most consensus is subjective experience (“what it is like”), and the least consensus surrounds luminosity / self-luminous awareness and intrinsic fundamentality; nearly every criterion in between is a behavioural or functional correlate that thinkers agree is real but disagree on whether it touches experience itself.

Almost none of the ten criteria function as a proof of consciousness; they are beliefs and inferences, defeated by the problem of other minds, the hard problem, and edge cases such as dreams, blindsight, locked-in syndrome and contentless meditative states.

Any generic test for consciousness fails at the same joint every time: it can only ever measure third-person function, never first-person experience, which is why contemplative traditions such as Abhinavagupta’s Kashmir Shaivism reframe the whole exercise as a category error, treating consciousness as the ground that makes testing possible rather than an object to be tested.

Lead-in

We do not know what consciousness is. That is not a confession of laziness; it is the honest state of the field. The philosopher David Chalmers separated the “easy problems” (how the brain discriminates, integrates, reports) from the “hard problem” (why any of that processing is accompanied by experience at all), and the hard problem has not budged. When the 2020 PhilPapers survey asked philosophers whether there is a hard problem of consciousness, of the 1,092 respondents who answered, 62.86% accepted or leaned toward “yes” versus 28.29% “no” (Bourget and Chalmers, “Philosophers on Philosophy: The 2020 PhilPapers Survey”). Even our most confident theories measure the shadow consciousness casts, not the thing itself.

This uncertainty is not a reason for paralysis. It is a reason for care. If we cannot prove who or what has an inner life, then the morally serious response is to widen the circle of our concern rather than narrow it. Two maps help us hold this uncertainty without pretending to resolve it. The first is panpsychism, defended in recent analytic philosophy by Galen Strawson and Philip Goff, which treats experience as a fundamental feature of reality rather than a late arrival. The second is Abhinavagupta’s Kashmir Shaivism, which treats consciousness (Cit) as the self-luminous ground of all appearance, recognised rather than detected. And the working model of how to act well under such uncertainty already exists in law: in 2017 New Zealand passed the Te Awa Tupua (Whanganui River Claims Settlement) Act, whose section 14 declares the river “a legal person” with “all the rights, powers, duties, and liabilities of a legal person”, and whose section 12 defines Te Awa Tupua as “an indivisible and living whole, comprising the Whanganui River from the mountains to the sea, incorporating all its physical and metaphysical elements” — not because anyone proved the river feels, but because a community decided that a being it could not fully know was worth protecting as if it mattered. The list that follows is offered in that spirit.

The ordering principle

“Consensus” here is layered. Where hard data exists, I use it: the 2020 PhilPapers survey gives us a rough ladder of which beings philosophers attribute consciousness to, which in turn reveals which criteria command agreement. Otherwise I rank by how widely a criterion is accepted as at least necessary to consciousness, and how concrete and measurable it is. A rough rule emerges: the more behavioural and third-person a criterion is, the more consensus it commands and the less it actually touches experience; the more it targets experience itself, the more it fractures into disagreement. This is the central irony of consciousness studies.

The top ten criteria (most to least consensus)

1. Subjective experience / phenomenal consciousness (“what it is like”)

What it is: The felt, first-person quality of being a subject at all: that there is “something it is like” to be you, in Thomas Nagel’s famous phrase from his 1974 paper “What Is It Like to Be a Bat?”. Simplest example: The redness of red as you actually see it. Problems: It is the most agreed-upon mark of consciousness and the least verifiable in anyone but oneself. You cannot confirm it in another being without already assuming it, which makes it the heart of the hard problem and the problem of other minds. Proof or belief: A proof in the first person, but a belief about everyone and everything else.

2. Sentience / the capacity to feel (especially to suffer)

What it is: The capacity to have valenced experiences: pleasure and pain, comfort and distress. Simplest example: A dog yelping and withdrawing its paw from a flame. Problems: Behavioural pain responses can occur without felt pain (reflexes, nociception in anaesthetised bodies), and felt states can occur without behaviour (paralysis). This is the criterion that carries most ethical weight, which is exactly why over-attributing or under-attributing it both do harm. Proof or belief: A belief, grounded in inference from analogy.

3. Wakefulness / arousal

What it is: The global on/off state of being awake and responsive rather than in dreamless sleep, anaesthesia or coma. Simplest example: Opening your eyes and becoming alert in the morning. Problems: This is the most clinically measurable criterion, and precisely for that reason it is the most obviously a correlate rather than the thing itself. Dreaming is vivid consciousness during behavioural sleep; some anaesthetised patients have experiences; wakefulness and experience come apart. Proof or belief: A belief; a reliable correlate, not a proof of experience.

4. Responsiveness / reportability

What it is: The ability to respond to stimuli and, in the strong case, to report on one’s own mental states. Simplest example: Saying “yes, I can see the light” when a light is switched on. Problems: Reportability is the working criterion of most laboratory science, but it conflates access with experience. Ned Block’s distinction between access consciousness and phenomenal consciousness shows that report may capture only part of what is experienced. Locked-in patients are fully conscious yet cannot respond; blindsight patients respond accurately to stimuli they report not seeing. The Turing test is a test of this criterion, and Searle’s Chinese Room argues that fluent response can be pure syntax with no understanding behind it. Proof or belief: A belief; it tests function, not experience.

5. Intentionality / aboutness

What it is: The “directedness” of mental states at objects: a thought is always a thought of or about something. Franz Brentano called intentionality “the mark of the mental.” Simplest example: Thinking about your grandmother. Problems: Brentano’s thesis that intentionality is unique to the mental is widely rejected today; a thermostat or a word on a page can be said to be “about” something without anyone thinking it is conscious. And some conscious states (undirected anxiety, a wash of contentless awareness) seem to lack a clear object. Proof or belief: A belief; at best a frequent feature, not a definition.

6. Unity / integration of experience

What it is: The fact that experience comes bound together into a single unified field rather than as scattered fragments. Simplest example: Seeing the colour, hearing the crunch and tasting the sweetness of an apple as one experience. Problems: This criterion anchors Integrated Information Theory (IIT), which proposes consciousness is integrated information and is measurable as a quantity. But integration is a structural property; that a system integrates information does not explain why integration should feel like anything. Split-brain cases suggest unity can fracture while experience persists. Proof or belief: A belief; integration is a correlate that IIT elevates to a definition, contentiously.

7. Self-awareness / metacognition

What it is: Awareness of oneself as a subject, and the capacity to monitor one’s own mental states (“thinking about thinking”). Simplest example: Recognising yourself in a mirror. Problems: Self-awareness seems too demanding: infants, many animals and people in absorbed states are surely conscious without robust self-reflection. Higher-order theories that build consciousness from self-monitoring face the charge of circularity, and contemplative traditions insist self-awareness can exist without a represented “self” at all. Proof or belief: A belief; it conflates a high-level capacity with experience as such.

8. Agency / volition

What it is: The capacity to initiate action, to choose, to act for reasons rather than merely react. Simplest example: Deciding to raise your hand, and raising it. Problems: Agency is observed from the outside and inferred, and sophisticated goal-directed behaviour in simple organisms and machines shows action without any obvious inner life. Conversely, a fully conscious but paralysed person has experience without agency. Proof or belief: A belief; behaviour underdetermines experience.

9. Temporality / the stream of consciousness

What it is: The flowing, continuous character of experience through time, William James’s “stream of consciousness,” and its connection to memory and anticipation. Simplest example: Hearing a melody as a flowing tune rather than disconnected notes. Problems: Stream-like continuity is a feature of normal waking experience but may be absent in certain meditative or pathological states, and its felt continuity may itself be a construction. Proof or belief: A belief; a description of structure, not a test.

10. Luminosity / self-luminous awareness

What it is: The contemplative claim that consciousness is intrinsically “lit up”, self-revealing, knowing itself without needing a second observer. In Advaita Vedanta it is svayam-prakasha (self-luminous) witness-awareness (sakshi); in Buddhism the mind is prabhasvara (luminous, clear light); in Kashmir Shaivism it is prakasha (light) inseparable from vimarsha (self-reflective awareness). Simplest example: The fact that you do not need a second light to know you are aware; awareness already shows itself. Problems: From a third-person scientific standpoint this is the least testable criterion of all, accessible only through trained first-person contemplative practice. Yet it is precisely the criterion that exposes the others, because it claims that consciousness is never an object and so can never appear in any test. Proof or belief: Held as a direct realisation within contemplative traditions, but a belief from any third-person standpoint.

A step-by-step attempt to build a test for consciousness (and why each step fails)

Imagine we want a general test we could run on anything: a human, an animal, a plant, a river, any system at all. Here is the honest sequence.

Step 1: Look for behaviour. Does the thing respond to its environment, avoid damage, pursue goals? Why it fails: behaviour is function, not experience. Searle’s Chinese Room shows that perfectly appropriate responses can be generated with no understanding inside. This step measures the “easy problems” and leaves the hard one untouched.

Step 2: Look for report. Ask the thing about its inner states. Why it fails: most candidates cannot report at all (plants, rivers, infants), so absence of report cannot mean absence of experience, or locked-in patients would count as unconscious. And report is just a special kind of behaviour, vulnerable to the same gap.

Step 3: Look inside at the mechanism. Scan for the neural correlates, the global workspace, the integrated information. Why it fails: a correlate is a correlation, not an identity. Even if every conscious state in humans co-occurs with brain pattern X, we have not shown X is experience or produces it.

Step 4: Argue by analogy. It resembles me; I am conscious; therefore it probably is too. Why it fails: analogy weakens fast as the candidate grows less like us. The 2020 PhilPapers “Other minds” data show exactly this descent: adult humans 95.1%, cats 88.6%, fish 65.3%, flies 34.5%, worms 24.2%, plants 7.2%, particles 2.0%, with the sharpest break falling between fish and worms. Analogy yields graded belief, never proof.

Step 5: Try first-person verification. Just be the thing and check. Why it fails: you cannot. The one case where you have direct access (your own experience) is the one case you did not need a test for, and it is unrepeatable and unshareable. This is the wall.

The verdict: every step either measures function (and misses experience) or relies on an unprovable assumption. The test cannot be completed. The problem of other minds is not a gap in our instruments; it is structural. We are always inferring, never confirming.

The contemplative reframing: Abhinavagupta and the recognition of what already is

Here the contemplative traditions, and Kashmir Shaivism in particular, make a move that turns the whole project inside out. Their objection is not that our test is too crude. It is that we have mistaken the ground for an object.

In the Pratyabhijñā (“Recognition”) school of Utpaladeva (c. 925–950 CE) and his great commentator Abhinavagupta (c. 975–1025 CE), consciousness (cit, samvit) is svaprakasha, self-luminous: it is that by which everything else is made manifest, and so it can never itself appear as one more manifested object. As Isabelle Ratié reconstructs the argument in “The Dreamer and the Yogin” (Bulletin of SOAS, 2010), the other’s consciousness cannot be perceived as an object precisely because its nature is to be self-manifest (svaprakasha), contrary to objects, which require a consciousness in order to be manifested. A test, by definition, detects objects within awareness. Awareness is the precondition of any detection, not an item that could ever turn up in the results. The knower cannot be made the known.

The point is sharpened by the school’s idealist argument: the very attempt to conceive of something wholly “alien to consciousness” is incoherent, because, in Ratié’s summary of Utpaladeva, “an object by nature alien to consciousness is simply unthinkable.” Every test already presupposes the consciousness it claims to be searching for.

The system’s positive structure is the dyad of prakasha and vimarsha. Prakasha is the sheer light of consciousness, its self-revealing shining; vimarsha is that light’s reflexive awareness of itself, the “I” knowing itself as light. Without vimarsha, the tradition holds, light would be mere inert radiance; a diamond shines but does not know it shines. Consciousness is the light that knows itself shining. This is why the path is called pratyabhijna, re-cognition: liberation is not the acquisition of a new fact or the detection of a hidden property, but the recognition of what one has always already been.

So Kashmir Shaivism would gently decline the entire list-making exercise, not by adding an eleventh criterion but by pointing out that the list-maker is the very consciousness being looked for. Detection presupposes a detector; the search for consciousness is consciousness looking for itself and failing to recognise its own face. And it returns us to the theme: if consciousness is the ground rather than an object, then we will never get the proof we want, and the only responsible posture under that permanent uncertainty is care — the stance the Whanganui settlement models in law: to treat what we cannot fully know as if it might matter, because it might.

Caveats

• The PhilPapers figures describe Anglophone academic philosophers, a specific and unrepresentative population; they capture the state of a debate, not the truth of the matter. (For context, the same survey found physicalism about the mind at 51.9% and non-physicalism at 32.1%.)
• Neuroscientific theories (IIT, Global Workspace) are described here only as they bear on definition; the report deliberately does not adjudicate the empirical neuroscience, and notes throughout that such theories measure correlates of consciousness rather than experience itself.
• The ordering by “consensus” is a reasoned synthesis, not a measured quantity; reasonable thinkers would shuffle the middle ranks, though the top two and the contemplative tenth are robust.
• Contemplative claims (luminosity, self-luminous awareness, recognition) are reported faithfully as first-person traditions hold them; their truth is not assumed, and the report flags that contentless awareness is itself contested.