77

What Is Wrong with the No-Report Paradigm and How to Fix It Ned Block1,* Is consciousness based in prefrontal circuits involved in cognitive processes like thought, reasoning,andmemoryorisitbasedinsensoryareasinthebackoftheneocortex?Theno-report paradigmhasbeencrucialtothisdebatebecauseitaimstoseparatetheneuralbasisofthecognitive processes underlying post-perceptual decision andreportfromtheneural basisofconscious perception itself. However, the no-report paradigm is problematic because, even in the absence of report, subjects might engage in post-perceptual cognitive processing. Therefore, to isolate theneuralbasisofconsciousness,ano-cognitionparadigmisneeded.Here,Idescribeano-cognition approach to binocular rivalry and outline how this approach can help to resolve debates about the neural basis of consciousness. What Is the Neural Basis of Consciousness? In recent years the scientific study of consciousness (see Glossary)hasfocusedonfindingtheneural basis of consciousness in the brain. There are manytheoriesoftheneural basis of consciousness, but in broadstrokestheories tend todivideonwhetherconsciousnessisrootedinthe‘front’orthe‘back’ of the brain. More specifically, they divide on whether perceptual consciousness is based in the neural circuits in the prefrontal and parietal cortex that are devoted to cognitive processes like thinking, reasoning, evaluating, reporting, deciding, and memory or whether perceptual consciousness is based in areas in the occipital and temporal cortex that are devoted to sensory processing. Garnering firm support for cognitive theories versus sensory (or noncognitivist) theories has proved methodologically challenging. However, it is widely held that the no-report paradigm [1] can help toadjudicatebetweenthesetwoperspectives.Thelogicoftheno-reportparadigmisasfollows.Suppose weseek to isolate the neural basis of consciously seeing, say, a face. Clearly, we would want to contrast brain activations in which a subject is consciously seeing a facewithbrainactivationsinwhich the subject is not consciously seeing a face. However, we typically know whether subjects are consciously seeing a face only via their own self-reports. So, it may seem that the neural basis of consciousness ofafacewill inevitablybeentangledwiththeneuralbasisofthepost-perceptualcognitive processes underlying judging what the stimulus is, maintaining that answer in working memory, and deciding what response to make [1,2]. The no-report paradigm is supposed to be a way of finessing this problem by using reports to find andvalidate indicators of conscious perception that work in the absence of reports. Although the no-report paradigm has been influential in recent years, it has a fatal flaw: eliminating reports does not eliminate post-perceptual cognitive processes if the subjects are thinking and reasoning about the stimulus. We must find a way to replace the ‘no-report’ paradigm with a ‘nopost-perceptual cognition’ paradigm. In this opinion article, I suggest a path forward. The Opposing Sides As indicated above, theories tend to divide on whether consciousness is rooted in the front or the back of the brain. However, ‘front’ and ’back’ are vague terms, and both sides regard parts of the parietal cortex as part of the neural basis of consciousness. The real neocortex-location issue is whether certain regions (not all) in front of the central sulcus are necessary for perceptual consciousness. Prefrontalists say yes, advocates of the back say no. Prefrontalists emphasize dorsolateral, medial prefrontal, anterior cingulate, and orbitofrontal regions of the prefrontal cortex. However, front/back is really a surrogate for the more fundamental issue of whether consciousness is cognitive or whether it is perceptual (in a wide sense that includes imagery, dreaming, and hallucination). It is Highlights Cognitive approaches to consciousness dictate that consciousness involves frontal and parietal circuits that are devoted to thinking, reasoning, evaluating, reporting, deciding, and memory. By contrast, sensory approaches allowforconsciousnessincreatures that have little or no ability to think and reason. To decide between cognitive and sensory accounts, we must distinguish between the neural basis of consciousness and the neural basis of reports. The no-report paradigm purports to solve this problem by using the reports of some subjects to calibrate indicators of consciousness, allowing experimental subjects to make no report. The problem with the no-report paradigm is that you cannot keep subjects from thinking, and their thought processes may be stimulated by and reflect the contents of perception. Thesolution isa ‘no-post-perceptual cognition’ paradigm, one version of which is illustrated here. 1New York University, New York, NY, USA *Correspondence: ned.block@nyu.edu Trends in Cognitive Sciences, December 2019, Vol. 23, No. 12 https://doi.org/10.1016/j.tics.2019.10.001 1003 ª2019 Elsevier Ltd. All rights reserved.Trends in Cognitive Sciences because activations in some areas of the prefrontal cortex appear to be necessary for cognitive function that it looms so large in these debates. (What is meant by cognition here is thought-based mentality; notably, reasoning, deciding, evaluating, reporting, and working memory). Advocates of the ‘back of thehead’viewofconsciousness can allowthat if thought has its ownkind of phenomenal consciousness, itmight bebased incognitive regions of theprefrontal cortex. The important difference between cognitivist and noncognitivist views is that the noncognitivist says perceptual consciousness does not require cognitive processing. This ‘front of the head’ versus ‘back of the head’ debate maps onto popular theories of consciousness. Prefrontalists are represented by the global workspace theory [3] and the higher-order theory [4,5] of consciousness. According to the global workspace theory, sensory activations compete among themselves, with dominant neural coalitions in sensory areas triggering workspace neurons in the frontal and parietal cortex, forming an active reverberating network that makes sensory informationavailabletoreasoning,reporting, decision-making, andothercognitiveprocesses.According to higher-order theories of consciousness, what makes a perception conscious is that there is an accompanying cognitive state about the perception. Meanwhile, the back of the head accounts are represented by the integrated information theory (IIT) [6] and the recurrent activation theory [7] of consciousness. According to the IIT, a system is conscious to the extent that it is both differentiated and integrated. According to the recurrent processing account, consciousness is a matter of the formation of feedback loops with certain neural properties. This debate between advocates of the front and the back is widely recognized to be the focus of current thinkingaboutconsciousness.TheTempletonWorldCharityFoundationisfundingan‘adversarial collaboration’ to– among other things– resolve the issue between advocates of the front and the back of the neocortex [8]. The frontalists also recently published a manifesto [9] in Science, arguing that if we are to make conscious machines, we should base them on the computations that underlie human consciousness. The computations they describe are those advocated by global workspace and higher-order theorists. One caution about the neuroscientific work to be presented: it is all correlational and of course one must be careful in inferring anything about the neural basis of mentality from correlational studies. Causal studies in which the effects of brain interventions (e.g., lesions, electrical stimulation, optogenetics) are assessed are better to the extent that they are available. Using Binocular Rivalry to Study Consciousness Unusual phenomena often provide the test cases for theories. One such phenomenon is binocular rivalry, in which different stimuli presented to the two eyes result in oscillating perceptions. Binocular rivalry has playedacentralroleinconsciousnessresearchbecauseitallowsresearcherstoholdastimulus constant whilethecontentsofconsciousnessshift. Figure1Adepictsthebrainofasubjectwhois wearing red/green glasses and is viewing a red house superimposed on a green face. One eye receivesafacestimulusandtheotherahousestimulus.Theconsciousperception–showninFigure1Bis not a combined image but rather conscious alternation, with all or most of the visual field filled by the conscious perception of either a face or a house, alternating every few seconds. Subjects are aware of intermediate mixtures between a face and house, but the experience comprises mainly oneperceptortheother,withmorecompletedominanceifthestimuliaresmall.Subjectscaninfluence which of the percepts dominates, but inevitably the nondominant percept takes over. Whydoesthisalternation occur? According to the prevailing account of rivalry, pools of neurons representing each of the incompatible stimuli inhibit one another [10]. In the presence of neural noise, one pool wins temporarily. Then that pool is weakened by adaptation and the other pool representing the other alternative takes over. Because of the impact of neural noise, the time of the transitions cannot be predicted on the basis of past transitions. Binocular rivalry occurs in many animals, including fruit flies, and can occur in humans with invisible stimuli, showing that binocular rivalry is not intrinsically a conscious process [11]. Glossary Binocular rivalry: in binocular rivalry, ‘incompatible’ stimuli presented to each eye result in the dominance of first one stimulus, thentheother,adinfinitum.(What makes stimuli incompatible is described in Box 1.) Cognition: thought-based mental states and processes; for example, reasoning, deciding, evaluating, reporting, and memory. What is especially important to cognition is transitions among propositional states that are based on the contents of those states. Consciousness: phenomenal consciousness is what it is like to have an experience. Phenomenal consciousness can be distinguished at least at the conceptual level from access consciousnessthe global availability of information. In this opinion article, the term ‘consciousness’ is restricted to phenomenal consciousness. Fixation: to fixate a thing or area ofspaceistopointyoureyesatit. Global workspace theory of consciousness: according to the global workspace theory, sensory activations compete with one another for dominance. Dominant sensory neural coalitions trigger workspace neurons in the frontal and parietal cortex, forming an active reverberating network that makes sensory information available to reasoning, reporting, decision-making, and other cognitive processes. Higher-order theories of consciousness: according to higherorder theories of consciousness, what makes a perception consciousisthatitisaccompanied by a cognitive state about the perception. Higher-order theories of consciousness differ in whether the higher-order state itself has a sensory content that could compete with the first-order content or whether the higher-order content is more of a pointer with an index of the reliability of the f irst-order state. Integrated information theory (IIT) of consciousness: according to IIT, a system is conscious to the extent that it is differentiated and that its different possible states areintegratedwithoneanother.A highly conscious and therefore highly integrated system cannot 1004 Trends in Cognitive Sciences, December 2019, Vol. 23, No. 12Trends in Cognitive Sciences Smulus Yoked Smulus Percept (B) Rivalry (C) Nonrivalry Time (s) (A) Figure 1. Binocular Rivalry. (A) Thebrainofapersonlookingthroughredandgreenglassesatasuperimposedpictureofafaceandahouse.(B) The percept, first of a face, then a house, then a face. (C) The replay that in many experiments is compared with rivalry. I am grateful to Frank Tong for this diagram. See also [31]. be decomposed into separate subsystemsthatarethemselvesas highly integrated and differentiated. IIT is a theory of what makes a system a conscious subject, whereas the other theories of consciousness are primarily concerned with conscious states of a subject. Neural basis of consciousness: the minimal neural activity that is sufficient for consciousness. It is possible that there are different neural bases of consciousness in different conscious beings. No-report paradigm: inanoreport paradigm, subjects are assigned no task of discriminating one stimulus from another. In noreport paradigms, reports can be used to calibrate other methods of indexing consciousness, such as the OKN reported in the text. Optokinetic nystagmus (OKN): during a conscious percept of a moving grating, a subject’s eyes move slowly in the direction of motion and then sharply back in the opposite direction. Perception: sensory objective representation of the environment. Perceptual consciousness: conscious phenomenology involved in perception. What cognitivists and noncognitivists disagree about is whether perceptual consciousness requires prefrontal cortex activations. Noncognitivists allow that the nonperceptual phenomenologyofthoughtmightberootedin prefrontal cortex. Prefrontalists:thosewhohold that the activation of certain circuits in front ofthecentralsulcusis necessary for perceptual consciousness; notably, the dorsolateral, medial prefrontal, anterior cingulate, and orbitofrontal regions. Prefrontalists subscribe to either the global workspace theory or the higher-order theory of consciousness. Recurrent activation theory of consciousness: the content of consciousness depends on which circuits are activated, but what makesthosecontentsconsciousis reverberating feedback to early sensory areas that satisfies certain spatial and temporal constraints, especially if it involves burst firing of neurons. Trends in Cognitive Sciences, December 2019, Vol. 23, No. 12 1005Trends in Cognitive Sciences Decoding Transitions versus Decoding Contents Whensubjects in abinocular rivalry experiment are scanned during the rivalrous perceptions, what is found is neural correlates of the transitions between one percept and the other in visual areas in the back of the head and stronger differential activations in cognitive areas in the prefrontal cortex and parietal cortex. These experiments also isolated the neural basis of the perceptual contents themselves, activations in the fusiform face area in the case of face perceptions and activations in the parahippocampal place area in the case of house perceptions. Both areas are in the back of the head. A conclusion often drawn from these early rivalry experiments, especially on the basis of the stronger frontal correlations with transitions, was that although the neural basis of the contents of perception such as face content or house content was in the back of the head, what makes those contents conscious was based in the front of the head [12]. It is important to distinguish the neural basis of transitions in rivalry from the neural basis of the contents of the rivalrous states themselves [13]. Of course, what we are interested in is mainly the neural basis of the contents, but the neural basis of the transitions can be important in finding the neural basis of the contents. If there is no prefrontal difference linked to perceptual transitions, we can conclude that the perceptual contents cannot be prefrontal. However, if there is a prefrontal differencelinked toperceptual transitions, it is much less clear what to conclude. The prefrontal difference might be due to differences in preconscious stages of processing [14] or to differential attention to the changing stimulus, to elevation of arousal due to a transition, or to motor control of response keys rather than changes in the content of conscious perception. What is crucial in deciding between theories of consciousness is a method of separating out those transition-related changes that are systematically related to perceptual contents. Themethodologyofmanybinocularrivalryexperimentsisdesignedonlytodetecttransitions,notthe contents themselves. Often studies of binocular rivalry involve comparisons between binocular shifts and what is called ‘replay’– real changes between, say, a face stimulus and a house stimulus with the same contents. fMRI always involves subtraction of one condition from another (Figure 1C). In this paradigm, transition-related activations during replay are subtracted from transition-related activations during rivalry. Since the two cases are supposed to involve the same conscious perceptual contents, in principle what is left after the subtraction is correlates of the transitions without any information about the perceptual contents themselves.