Artificial intelligence, human cognition, and conscious supremacy.

The computational significance of consciousness is an important and potentially more tractable research theme than the hard problem of consciousness, as one could look at the correlation of consciousness and computational capacities through, e.g., algorithmic or complexity analyses. In the literature, consciousness is defined as what it is like to be an agent (i.e., a human or a bat), with phenomenal properties, such as qualia, intentionality, and self-awareness. The absence of these properties would be termed "unconscious." The recent success of large language models (LLMs), such as ChatGPT, has raised new questions about the computational significance of human conscious processing. Although instances from biological systems would typically suggest a robust correlation between intelligence and consciousness, certain states of consciousness seem to exist without manifest existence of intelligence. On the other hand, AI systems seem to exhibit intelligence without consciousness. These instances seem to suggest possible dissociations between consciousness and intelligence in natural and artificial systems. Here, I review some salient ideas about the computational significance of human conscious processes and identify several cognitive domains potentially unique to consciousness, such as flexible attention modulation, robust handling of new contexts, choice and decision making, cognition reflecting a wide spectrum of sensory information in an integrated manner, and finally embodied cognition, which might involve unconscious processes as well. Compared to such cognitive tasks, characterized by flexible and ad hoc judgments and choices, adequately acquired knowledge and skills are typically processed unconsciously in humans, consistent with the view that computation exhibited by LLMs, which are pretrained on a large dataset, could in principle be processed without consciousness, although conversations in humans are typically done consciously, with awareness of auditory qualia as well as the semantics of what are being said. I discuss the theoretically and practically important issue of separating computations, which need to be conducted consciously from those which could be done unconsciously, in areas, such as perception, language, and driving. I propose conscious supremacy as a concept analogous to quantum supremacy, which would help identify computations possibly unique to consciousness in biologically practical time and resource limits. I explore possible mechanisms supporting the hypothetical conscious supremacy. Finally, I discuss the relevance of issues covered here for AI alignment, where computations of AI and humans need to be aligned.

How and When? Metacognition and Solution Timing Characterize an "Aha" Experience of Object Recognition in Hidden Figures.

The metacognitive feelings of an "aha!" experience are key to comprehending human subjective experience. However, behavioral characteristics of this introspective cognition are not well known. An aha experience sometimes occurs when one gains a solution abruptly in problem solving, a subjective experience that subserves the conscious perception of an insight. We experimentally induced an aha experience in a hidden object recognition task, and analyzed whether this aha experience was associated with metacognitive judgments and behavioral features. We used an adaptation of Mooney images, i.e., morphing between a grayscale image and its binarised image in 100 steps, to investigate the phenomenology associated with insight: aha experience, confidence, suddenness, and pleasure. Here we show that insight solutions are more accurate than non-insight solutions. As metacognitive judgments, participants' confidence in the correctness of their solution is higher in insight than non-insight problem solving. Intensities of the aha feeling are positively correlated with subjective rating scores of both suddenness and pleasure, features that show marked signs of unexpected positive emotions. The strength of the aha experience is also positively correlated with response times from the onset of presentation until finding the solution, or with task difficulty only if the solution confidence is high enough. Our findings provide metacognitive and temporal conditions for an aha experience, characterizing features distinct from those supporting non-aha experience.

Adult attachment style and lateral preferences in physical proximity.

Attachment scale is constructed from two components (anxiety and avoidance) effectively treated as providing salient measures in previous studies. Recent studies have suggested associations between sensitivities to physical warmth and anxiety scores of attachment scale. Some researchers also suggest that the degree of one's comfort with physical proximity depends on attachment styles, attributing differences to the number of oxytocin (a neuropeptide released by physical touch) receptors. Lateral preference is an important aspect of physical proximity, coupled with the lateralization of visual, emotional, and other cognitive systems. However, there are few studies investigating the relationship between attachment scale scores and one's lateral preferences in physical proximity. Here we surveyed the preferences of subjects regarding positional relations with their romantic partner in some daily situations, and examined the association with attachment scale score. Our results show that the existence or absence of partner correlates with different relations between attachment styles and subjects' awareness of lateral preferences. Lateral preferences in physical proximity may play an important role in attachment in adulthood.

Multiple processes in two-dimensional visual statistical learning.

Knowledge about the arrangement of visual elements is an important aspect of perception. This study investigates whether humans learn rules of two-dimensional abstract patterns (exemplars) generated from Reber's artificial grammar. The key question is whether the subjects can implicitly learn them without explicit instructions, and, if so, how they use the acquired knowledge to judge new patterns (probes) in relation to their finite experience of the exemplars. The analysis was conducted using dissimilarities among patterns, which are defined with n-gram probabilities and the Levenshtein distance. The results show that subjects are able to learn rules of two-dimensional visual patterns (exemplars) and make categorical judgment of probes based on knowledge of exemplar-based representation. Our analysis revealed that subjects' judgments of probes were related to the degree of dissimilarities between the probes and exemplars. The result suggests the coexistence of configural and element-based processing in exemplar-based representations. Exemplar-based representation was preferred to prototypical representation through tasks requiring discrimination, recognition and working memory. Relations of the studied judgment processes to the neural basis are discussed. We conclude that knowledge of a finite experience of two-dimensional visual patterns would be crystalized in different levels of relations among visual elements.

Perception of temporal duration affected by automatic and controlled movements.

Sensorimotor contingency is one of the main factors to warp time perception. Voluntary actions such as saccades and hand movements affect the subjective perception of temporal duration. Although the perceived timings of action and stimulus are affected by whether an action was automatic or controlled, its effect on the subjective perception of duration has not been studied except in the case of saccade (chronostasis), which has been shown to be unaffected by the context of action initiation. Here we investigate the effect of the context of action initiation on duration estimation in the case of finger movement. The reproduced intervals were shorter when actions were initiated by automatic manner, compared to self-timed or cognitively controlled actions. The results are compatible with an internal clock model employing variable latencies for switch closure after action.

Free will and paranormal beliefs.

Free will is one of the fundamental aspects of human cognition. In the context of cognitive neuroscience, various experiments on time perception, sensorimotor coordination, and agency suggest the possibility that it is a robust illusion (a feeling independent of actual causal relationship with actions) constructed by neural mechanisms. Humans are known to suffer from various cognitive biases and failures, and the sense of free will might be one of them. Here I report a positive correlation between the belief in free will and paranormal beliefs (UFO, reincarnation, astrology, and psi). Web questionnaires involving 2076 subjects (978 males, 1087 females, and 11 other genders) were conducted, which revealed significant positive correlations between belief in free will (theory and practice) and paranormal beliefs. There was no significant correlation between belief in free will and knowledge in paranormal phenomena. Paranormal belief scores for females were significantly higher than those for males, with corresponding significant (albeit weaker) difference in belief in free will. These results are consistent with the view that free will is an illusion which shares common cognitive elements with paranormal beliefs.

Cognitive factors correlating with the metacognition of the phenomenal properties of experience.

The awareness of the phenomenal qualities of one's experiences can be considered as an instance of metacognition. Although some people take qualia (sensory qualities such as the redness of red) as salient features of phenomenal experience, others have expressed views that doubt or deny the central importance of qualia. How do such cognitive heterogeneities occur? What parameters influence them? Here I examine the relationship between the awareness of the phenomenal qualities of subjective experience (qualia and free will) and general cognitive tendencies. The awareness of qualia was found to be more varied among subjects compared to the belief in free will. Various cognitive tendencies correlated with the metacognition of phenomenal experience. The awareness of qualia was found to increase significantly with age, suggesting a continuous learning process. These results suggest that heterogeneities in the metacognition of phenomenal properties of experience are important constraints in human cognition.

Egocentric mental transformation of self: effects of spatial relationship in mirror-image and anatomic imitations.

Assessing the mental state of others by considering their perspective plays an important part in social communication. Imitation based on visual information represents a typical case of the translation of sensory input into action. Although humans are often successful in imitating complex actions, the mechanisms that underlie successful imitation are poorly understood. In earlier findings, it has been suggested that understanding others' minds through imitation is realized in the course of the comparison between the representations of the self and others, involving a transformation of the egocentric perspective to the allocentric one. There are two possible strategies of transformation between the representation of the self and others. One possible scenario is that the imitator perceives and imitates others as if looking in a mirror (mirror-image imitation, where, for example, the demonstrator's right hand corresponds to the imitator's left hand). Alternatively, the imitator might estimate the demonstrator's action using the anatomically congruent limb (anatomic imitation, where, for example, the demonstrator's right hand corresponds to the imitator's right hand). Here, we conducted a series of experiments in which the subjects imitated simple hand actions such as pushing a button presented from several different spatial orientations rotated at various angles. We observed that the imitators changed their strategy of imitation (mirror-image or anatomic imitation) depending on the nature of spatial configurations. Behavioral data from this study support the hypothesis that mirror-image and anatomic imitations provide complementary systems for understanding the actions and intentions of others.

Visual one-shot learning as an 'anti-camouflage device': a novel morphing paradigm.

Once people perceive what is in the hidden figure such as Dallenbach's cow and Dalmatian, they seldom seem to come back to the previous state when they were ignorant of the answer. This special type of learning process can be accomplished in a short time, with the effect of learning lasting for a long time (visual one-shot learning). Although it is an intriguing cognitive phenomenon, the lack of the control of difficulty of stimuli presented has been a problem in research. Here we propose a novel paradigm to create new hidden figures systematically by using a morphing technique. Through gradual changes from a blurred and binarized two-tone image to a blurred grayscale image of the original photograph including objects in a natural scene, spontaneous one-shot learning can occur at a certain stage of morphing when a sufficient amount of information is restored to the degraded image. A negative correlation between confidence levels and reaction times is observed, giving support to the fluency theory of one-shot learning. The correlation between confidence ratings and correct recognition rates indicates that participants had an accurate introspective ability (metacognition). The learning effect could be tested later by verifying whether or not the target object was recognized quicker in the second exposure. The present method opens a way for a systematic production of "good" hidden figures, which can be used to demystify the nature of visual one-shot learning.

Analysis of ongoing dynamics in neural networks.

Spontaneous neural activities in the cerebral cortex exhibit complex spatio-temporal patterns in the absence of sensory inputs [Arieli, A., Shoham, D., Hildesheim, R., Grinvald, A., 1995. Coherent spatio-temporal patterns of ongoing activity revealed by real-time optical imaging coupled with single-unit recording in the cat visual cortex. J. Neurophysiol. 73, 2072-2093; Arieli, A., Sterkin, A., Grinvald, A., Aertsen, A., 1996. Dynamics of ongoing activity: explanation of the large variability in evoked cortical responses. Science 273, 1868-1871], wandering among the intrinsic set of cortical states [Tsodyks, M., Kenet, T., Grinvald, A., Arieli, A., 1999. Linking spontaneous activity of single cortical neurons and the underlying functional architecture. Science 286, 1943-1946; Kenet, T., Bibitchkov, D., Tsodyks, M., Grinvald, A., Arieli, A., 2003. Spontaneously emerging cortical representations of visual attributes. Nature 425, 954-956]. Elucidating the nature of such spontaneous activities is one of the most intriguing challenges in the effort to understand the computational principles employed by the brain. The precise mechanism behind these salient phenomena, however, is not known. Here we model the ongoing dynamics of generic neural networks with attractor states using a conductance-based neuron model. Our realistic modeling shows the existence of up-states and down-states in the membrane potential, where the up-states exist as spatially clustered patches moving within the network. Our analysis shows that up-states are sustained by the balance between excitatory and inhibitory inputs. Synaptic depression and depolarization-dependent potassium channels can cause the transitions from the up-states to down-states by affecting the dynamics in differential manners. The velocity of patches depends on the firing frequency of excitatory neurons affected by contributing factors. These results suggest that the switching dynamics can be produced by the interactions within the local network, revealing the constraints on the nature of autonomous dynamics within the cortex.

Distinct neural processes of bodily awareness in crossed fingers illusion.

The tactile reassignment process supports the flexible and dynamic changes of body schema in various situations such as those involving tool use. Here, we show that there exist two distinct neural processes in the dynamical reassignment process. One process is involved in identifying the body part where the tactile stimuli are applied, whereas the other is involved in the assignment of the tactile stimuli in the external space including one's body. These processes, combined together, would facilitate the quick and appropriate acquisition of information from the environment, resulting in the speedy spatial perception and execution of motor activities. In addition, we show that the body posture affects the accuracy of tactile localization in the crossed fingers illusion.

Spatio-temporal dynamics of the visual system revealed in binocular rivalry.

From the evolutionary viewpoint, animals need to monitor the surrounding environment and capture salient features, such as motion, for survival. The visual system is highly developed for monitoring a wide area of visual field and capturing such salient features. In humans and primates, there is a wide binocular field, suggesting a necessity of integrating the images from the two eyes. Binocular rivalry [R. Blake, A neural theory of binocular rivalry, Psychol. Rev. 96 (1989) 145-167; R. Blake, N.K. Logothetis, Visual competition, Nat. Rev. Neurosci. 3 (2002) 13-21], where incompatible inputs from the two eyes compete to emerge in the subject's visual percept, has been shown to exhibit highly adaptive behavior [I. Kovacs, T.V. Parathomas, M. Yang, A. Feher, When the brain changes its mind: interocular grouping during binocular rivalry. Proc. Natl. Acad. Sci. U.S.A. 93 (1996) 15508-15511; N.K. Logothetis, Single units and conscious vision, Philos. Trans. R. Soc. Lond. B. Biol. Sci. 353 (1998) 1801-1818]. Here we investigated the spatio-temporal dynamics of the ocular dominance pattern in binocular rivalry under conditions where conflicting salient features were presented in a temporally varying manner. We found a striking example of the detailed structure of the dominance wave propagation, by using a spatio-temporal sampling method. The data show in detail the ability of the visual system to dynamically adapt to the changing stimuli in the context of the massively parallel visual field. We show by model prediction that the globally coherent dominance change in the presence of multiple stimuli can be explained by a mechanism based on local saliency comparison.

Dynamics of betting behavior under flat reward condition.

One of the missions of the cognitive process of animals, including humans, is to make reasonable judgments and decisions in the presence of uncertainty. The balance between exploration and exploitation investigated in the reinforcement-learning paradigm is one of the key factors in this process. Recently, following the pioneering work in behavioral economics, growing attention has been directed to human behaviors exhibiting deviations from the simple maximization of external reward. Here we study the dynamics of betting behavior in a simple game, where the probability of reward and the magnitude of reward are designed to give a "zero" expected net reward ("flat reward condition"). No matter how the subject behaves, there is on average no change in one's resources, and therefore every possible sequence of action has the same value. Even in such a situation, the subjects are found to behave not in a random manner, but in ways showing characteristic tendencies, reflecting the dynamics of brain's reward system. Our results suggest that brain's reward system is characterized by a rich and complex dynamics only loosely coupled with external reward structure.