In 1995, the philosopher David Chalmers published Facing Up to the Problem of Consciousness, distinguishing what he called the easy problems of consciousness (explaining cognitive functions: discrimination, reportability, attention, the difference between waking and sleeping) from the hard problem: why is there something it is like to undergo conscious experience at all? The easy problems are not easy in any absolute sense — they involve enormous neuroscientific and computational complexity — but they are easy in principle: they will yield to the ordinary methods of cognitive science given enough time and data. The hard problem is categorically different: even after every neural correlate of every conscious state has been identified, we will still not have explained why the neural activity is accompanied by subjective experience rather than nothing at all. The paper has been cited tens of thousands of times, and the problem has not been solved.
Consciousness — the phenomenal property of there being something it is like to undergo a mental state, what philosophers since Thomas Nagel (1974, What Is It Like to Be a Bat?) call qualia — is the last domain of human cognition for which no satisfactory scientific account exists. The hard problem distinguishes three aspects of conscious states: access consciousness (the information is available for reasoning, report, behavioural control), self-consciousness (a model of oneself is engaged), and phenomenal consciousness (there is something it is like to undergo this state) — the first two are easy problems, the third is hard. The major positions cluster: Type-A materialism (Dennett, Consciousness Explained, 1991) holds the hard problem is illusory; Type-B materialism identifies phenomenal consciousness with physical states but says the identity is brute and not deducible a priori; property dualism (Chalmers's 1995 view) holds that phenomenal properties are non-physical properties of physical systems; panpsychism (Chalmers, Strawson, Goff) holds phenomenal consciousness as fundamental and ubiquitous. Other frameworks add empirical traction: higher-order theories (Rosenthal, Carruthers) make a state conscious iff there is a higher-order representation of it; global workspace theory (Baars, Dehaene) offers a well-supported neural correlate of access consciousness; integrated information theory (IIT, Tononi 2004+) quantifies consciousness as φ (phi); predictive processing (Friston, Clark) traces consciousness to the brain's hierarchical predictive coding. Substantial empirical progress has been made on neural correlates of consciousness — the posterior hot zone of cortex, thalamocortical loops, gamma-frequency oscillations, global workspace ignition — and disorders of consciousness have been studied with fMRI to detect residual awareness, but none of this bridges to the hard problem: it identifies which neural states are accompanied by consciousness without explaining why.
The hard problem remains unresolved. The philosophical positions have firmed up but not converged — consensus exists that the problem is real (against Type-A) but not on which alternative is correct, with recent contributions including Annaka Harris's Conscious (2019), Philip Goff's Galileo's Error (2019, defending panpsychism), and Anil Seth's Being You (2021, predictive-processing approach). AI consciousness now forces the question afresh — do large language models have phenomenal experience? Most experts believe no, but the grounds for confidence are weak. Whether the hard problem can be solved at all is itself contested — mysterianism (Colin McGinn) holds that the human cognitive apparatus is constitutively unable to grasp the solution. The empirical neuroscience continues to advance, the philosophical situation has not changed in three decades, and whether this is a temporary or permanent stalemate is the open meta-question.