Abstract
The explanatory gap in consciousness research concerns the apparent distance between neural information processing and subjective experience. A particularly persistent version of the problem asks why a neural representation of a visual wavelength should produce the phenomenal experience of red rather than merely register, discriminate, or respond to red stimuli. The iterative updating model of working memory offers a possible response by treating conscious experience as the mode in which representational content exists when it is sustained within a continuously changing field of attention. According to this model, successive states of working memory overlap because some representations remain active while others are removed and replaced. Each state therefore incorporates part of its predecessor and contributes to the formation of its successor. This article develops the hypothesis that sensory and affective representations become phenomenally conscious when they participate in this state-spanning process. The sensory architecture producing a representation determines the qualitative character of the experience, while iterative integration supplies temporal extension, contextual embedding, global influence, and subjective presence. The experience of red can consequently be interpreted as a modality-specific visual representation undergoing sustained incorporation into an iteratively updated global workspace. This process, termed iterative phenomenalization, may account for the distinction between unconscious sensory registration and conscious perception without requiring a separate nonphysical property to be added to neural activity. The framework also generates predictions concerning masking, attentional selection, recurrent sensory activation, temporal integration, perceptual fading, and potential artificial consciousness.
Keywords: consciousness, qualia, phenomenal consciousness, working memory, iterative updating, attention, visual perception, color, global workspace, recurrent processing, artificial consciousness
Introduction
Scientific accounts of consciousness generally attempt to identify the neural and computational conditions under which information becomes available to awareness. Global workspace theories emphasize the widespread broadcasting of selected information across cognitive systems, recurrent-processing theories emphasize feedback interactions within sensory hierarchies, higher-order theories emphasize representations of mental states, and integrated information approaches emphasize the causal interdependence of system components. These frameworks have clarified many functional properties associated with consciousness, including reportability, access, integration, discrimination, memory formation, and behavioral control. Their relationship to phenomenal character remains disputed because a description of information processing can appear to leave unanswered why the processing should feel like anything from the perspective of the organism.
The experience of color offers a useful test case. Visual systems discriminate wavelength distributions, compare chromatic signals, separate objects from backgrounds, and guide behavior according to learned color associations. These capacities can be described in terms of neural coding and computation. The continuing philosophical question concerns why one of these representations is experienced as red. The representation appears to possess a subjective quality that cannot be captured by describing only its stimulus conditions, neural location, or behavioral effects.
The iterative updating model proposes that the contents of working memory are maintained through partially overlapping states. Some active representations are retained, some lose activity, and others enter the focus of attention. Successive states therefore share part of their informational content and form an incrementally changing sequence rather than a succession of independent configurations. This architecture has been proposed as a model of working memory, thought, cognitive continuity, associative search, and machine consciousness.
The present article extends this model into a stronger theory of phenomenal consciousness. It proposes that iterative overlap does more than connect conscious states after they have already become conscious. Iterative incorporation may be the process through which representational activity acquires subjective presence. On this account, the qualitative identity of an experience arises from the specialized representational system producing it, while its conscious status arises from its sustained participation in an iteratively updated field of attention.
From Sensory Registration to Conscious Presence
Sensory systems contain extensive unconscious processing. A visual stimulus may alter neural firing, prime later responses, guide eye movements, or influence decisions even when the observer reports no conscious perception. The presence of a discriminative neural signal is therefore insufficient by itself to establish phenomenal consciousness. A representation must enter a form of processing in which it persists, combines with other information, influences subsequent states, and becomes incorporated into the temporally extended condition of the organism.
Within the iterative updating model, the focus of attention is embedded within a broader short-term store. The focus contains highly active, consciously attended representations, while recently active information remains available at lower levels of prioritization. Information in these stores persists through different mechanisms and over different timescales. Sustained activity may maintain highly prioritized contents for seconds, while activity-silent changes in synaptic efficacy may preserve recent information for longer intervals.
When a representation enters the focus of attention, it is not simply placed inside a static register. It joins a changing collection of active representations that collectively determine the next update. Some elements of the present state persist into the next state, allowing recently acquired information to remain causally active as new sensory and internally generated content arrives. Consecutive states are therefore nested within one another through partially conserved activity. The model describes this as state-spanning coactivity and incremental change in state-spanning coactivity.
A sensory representation of red may initially arise through feedforward visual processing. If it remains weak, local, or rapidly extinguished, it may affect the system without becoming experienced. If the representation is selected, recurrently stabilized, integrated with object and spatial information, and carried through several partially overlapping attentional states, it becomes part of the organism’s temporally extended present. The hypothesis developed here is that this sustained incorporation constitutes its phenomenal presence.
The Representational and Temporal Dimensions of Qualia
A theory of qualia must explain both differentiation and presence. Differentiation concerns why red differs from green, pain differs from warmth, and sound differs from color. Presence concerns why any of these representational states is experienced rather than processed unconsciously.
The iterative account assigns these problems to related but distinguishable properties of neural organization. The qualitative identity of an experience depends on the representational architecture that produces it. Redness depends on the activity of visual systems shaped by retinal photoreceptors, opponent channels, cortical feature processing, spatial organization, perceptual categorization, and learned associations. Pain depends on nociceptive, interoceptive, affective, and action-related networks. Pitch depends on auditory frequency organization and temporal coding. The specific structure of the activated representation determines the content of experience.
Iterative incorporation determines whether that content becomes part of consciousness and how strongly it is present. A representation that remains active across several updates gains phenomenal duration. Its relationships with surrounding representations give it contextual meaning. Its participation in global processing allows it to influence memory, expectation, decision-making, imagery, verbal report, and action. Its residual effects help shape the interpretation of subsequent events.
Phenomenal character can therefore be conceptualized as a joint function:Qt=Φ(Rt,Ct,Ot)
Here, Rt denotes the modality-specific representation active at time t, Ct denotes the concurrent sensory, conceptual, affective, and motivational context, and Ot denotes the pattern of overlap connecting the representation to preceding and succeeding states. Qt denotes the phenomenal state generated by this combination.
This formulation does not imply that iterative overlap independently manufactures redness. The visual representation supplies chromatic content. Iterative integration gives that content persistence, contextual depth, and membership in the conscious field. A concise statement of the position is that representational structure determines what is experienced, while iterative temporal integration determines its conscious presence.
The Conscious Experience of Red
Seeing a red object involves more than activating an abstract category corresponding to the word red. The visual experience includes chromatic properties, spatial extension, boundaries, contrast, luminance, surface characteristics, object identity, expected persistence, and relations to surrounding features. The representation is also influenced by attention, memory, goals, affective significance, and anticipated action.
A red traffic signal, a red apple, a red afterimage, and an imagined red circle share chromatic organization while differing in their larger representational states. The representation of red is therefore neither an isolated feature nor an invariant local activation pattern. It is a context-sensitive neural ensemble whose exact composition changes according to the other contents active in working memory. The iterative updating model describes working-memory items as fuzzy ensembles assembled from lower-level neural components, with their active composition varying according to context.
The experience of red can be modeled as the persistence of a chromatic representation through an evolving series of states. At an initial moment, red-related sensory activity enters the attentional field together with representations of location, shape, brightness, and object identity. During the next update, some of these representations remain active while additional information is incorporated. The observer may recognize the object, retrieve an association, form an expectation, or prepare an action. Redness persists through these transformations, retaining its identity while participating in a changing perceptual and cognitive context.
This persistence gives the experience temporal thickness. The red percept belongs to a short interval containing traces of the immediately preceding state and anticipatory effects on the state that follows. A conscious percept is consequently better described as a trajectory through representational space than as a dimensionless neural instant. What the observer experiences as a stable patch of red may consist of a rapidly changing process whose successive configurations preserve sufficient representational overlap to support perceptual continuity.
Iterative Phenomenalization
The term iterative phenomenalization can be used to describe the process by which a sensory, affective, mnemonic, or conceptual representation becomes incorporated into subjective experience through sustained participation in overlapping attentional states.
Iterative phenomenalization contains several interdependent operations. A representation must first achieve sufficient activation to enter the focus of attention. It must remain active long enough to overlap with subsequent states. It must influence the selection or interpretation of new content. It must interact with other active representations in a common workspace. It must also remain connected to the modality-specific systems that supply its sensory or affective character.
The final requirement distinguishes conscious perception from purely abstract cognition. An internal token representing the proposition the object is red could guide reasoning without reproducing the phenomenal quality of seeing red. Conscious chromatic experience likely requires recurrent engagement with lower-level visual systems whose topographic and feature-specific organization supplies the perceptual structure of color.
The model therefore proposes a temporal incorporation criterion:
A representation becomes phenomenally conscious to the extent that it is integrated into state-spanning activity within an iteratively updated global workspace while remaining functionally coupled to the representational systems that determine its modality-specific character.
This criterion allows phenomenal consciousness to vary by degree. A weak representation may enter only briefly, exert minimal influence, and leave little residual activity. Such a representation may correspond to a vague, fleeting, or fringe experience. A strongly sustained representation may dominate several updates, recruit extensive associations, shape perception, and remain readily reportable. Such activity would correspond to vivid conscious presence.
A simplified relationship can be expressed as:P(x)∝I(x)D(x)G(x)R(x)
In this expression, P(x) is the phenomenal presence of representation x, I(x) is its integration with concurrent contents, D(x) is its persistence across updates, G(x) is the breadth of its influence on global processing, and R(x) is the strength of recurrent interaction with relevant sensory or affective systems.
The expression is not intended as a completed quantitative theory. It identifies dimensions that could eventually be operationalized and measured. Consciousness would increase as a representation becomes more strongly integrated, temporally sustained, globally influential, and recurrently connected to the neural architecture generating its qualitative content.
Phenomenal Texture as a Distribution of Temporal Traces
At any moment, the contents of consciousness possess different temporal histories. Some features have just entered awareness. Others have been active for several updates. Some are fading from the focus while remaining available in short-term memory. Goals, emotions, bodily states, perceptual expectations, and autobiographical concerns may persist longer than individual sensory details. Conscious experience therefore consists of a layered distribution of representational traces with different ages, strengths, and durations.
This distribution may account for what can be called phenomenal texture. An experience feels structured and substantial because it contains information arriving and decaying at multiple timescales. The present includes immediate sensory input, residual features from the recent past, stable contextual representations, active predictions, and preparations for future action. Each component contributes differently to the evolving state.
The iterative updating model proposes that partial replacement allows relevant information from the recent past to combine with present information, creating continuity across conscious states. This continuity should be understood as more than the preservation of an autobiographical narrative. It operates within individual percepts. A color patch remains perceptually present because its representation survives repeated updates despite microsaccades, changing sensory input, attentional fluctuations, and the continuous introduction of new information.
Phenomenal texture may therefore arise from structured persistence. The conscious field is neither static nor fully replaced from moment to moment. It is continually reconstructed from elements that occupy different positions along a trajectory of activation, maintenance, demotion, and reactivation.
Recurrent Sensory Processing and Progressive Perceptual Reconstruction
The iterative account gains explanatory power when combined with reciprocal processing between association areas and sensory systems. Higher-order representations can send top-down signals to visual, auditory, somatosensory, and interoceptive networks. Activity within these lower-level networks can then return to association areas and contribute to the next state of working memory.
The iterative updating model describes internally generated sensory maps as being progressively modified in synchrony with changing working-memory content. Consecutive images can preserve some constraints while altering others, producing a continuous sequence of related sensory constructions. The same reciprocal architecture can be applied to ordinary perception. Bottom-up sensory input constrains the active workspace, while top-down expectations and attentional priorities alter the interpretation and reconstruction of the sensory scene.
The experience of red may therefore arise from a recurrent loop rather than a unidirectional transfer of information into an amodal workspace. Visual activity contributes a chromatic representation to working memory. The updated workspace sends contextual constraints back into the visual hierarchy. The resulting sensory state then contributes to the next update. Across repeated cycles, the representation becomes stabilized, contextualized, and progressively elaborated.
This framework also clarifies differences among perceiving red, imagining red, remembering red, and thinking verbally about red. Perception is dominated by bottom-up sensory constraints. Imagery depends more strongly on top-down reconstruction. Memory may reactivate partial sensory patterns together with episodic context. Verbal thought may activate the abstract semantic representation while only weakly engaging visual maps. All four states can involve the concept of red, while their phenomenal qualities vary according to the strength, precision, and direction of recurrent sensory interaction.
The Explanatory Gap Reconsidered
The explanatory gap is often framed by asking why a neural signal associated with red should produce the experience of red. This formulation may presuppose that the signal and the experience are separate events requiring an additional causal bridge. A process-based identity theory rejects that presupposition. The organized neural process does not first occur and then generate a second, non-neural entity called the experience. The process, considered from the perspective of the system undergoing it, is the experience.
From a third-person perspective, the event can be described as recurrent activation of modality-specific visual representations integrated into a temporally overlapping global workspace. From a first-person perspective, the event is described as seeing red. The two descriptions refer to the same organized process at different explanatory levels.
A scientific account may identify the mechanisms that determine the structure, duration, integration, accessibility, and behavioral consequences of the experience. It cannot reproduce the experience merely by describing it. A verbal account of red does not activate the reader’s visual system in the same manner as direct perception. This limitation may reflect the difference between description and instantiation rather than the absence of an explanatory mechanism.
The proposed response to the explanatory gap can be stated directly:
Feeling is the mode in which modality-specific information exists when it is recurrently sustained, contextually integrated, and carried forward through an iteratively updated conscious workspace.
Under this interpretation, the felt quality of red is the intrinsic form of a chromatic representation embedded within the temporally extended activity constituting the subject’s conscious field. No additional phenomenal substance is required. The qualitative property corresponds to the representational organization, while subjectivity corresponds to the organization’s participation in an ongoing self-propagating process.
Some philosophers will regard this response as insufficient because it accepts a physicalist identity between neural organization and experience. The theory nevertheless advances the discussion by specifying which form of organization is proposed to constitute consciousness. It identifies partial temporal overlap, recurrent integration, global influence, modality-specific reentry, and progressive state transformation as central features of phenomenal processing.
Attention, Consciousness, and Graded Incorporation
The claim that everything in attention becomes conscious requires qualification because attention refers to several distinct phenomena. Neural processing can be enhanced or selected without producing a clearly reportable experience. A narrow definition of the focus of attention is more suitable for the iterative account.
The model proposes that consciously attended contents occupy the most active region of a broader working-memory system and interact with sensory input, current goals, recent context, and long-term memory. A representation becomes conscious when it is incorporated into this integrated, state-spanning process rather than merely receiving local gain enhancement.
Consciousness may therefore be graded according to the depth of incorporation. A stimulus may receive enough processing to influence behavior while failing to persist across multiple states. Another stimulus may enter the workspace briefly but remain weakly bound to other contents. A vivid percept may persist across many updates, dominate multiassociative search, reactivate sensory circuitry, and influence several cognitive and motor systems.
This graded view is compatible with phenomena such as partial awareness, uncertain perception, attentional blink, perceptual masking, fringe consciousness, and gradual fading. Each may involve differences in how long and how extensively a representation participates in the iterative stream.
Multiassociative Search and the Elaboration of Experience
In the iterative updating model, the active contents of working memory collectively influence which representation enters next. The current state functions as a set of search parameters, and spreading activation converges on contextually appropriate additions. Every update consequently alters the conditions that will determine the following update.
When red enters awareness, it contributes to associative search together with other active contents. A red object shaped like an apple may recruit representations related to food, taste, ripeness, personal memory, or intended action. A red light at an intersection may recruit stopping, danger, traffic rules, and braking behavior. The subjective character of the chromatic feature is preserved, while the larger experience develops through its interaction with the current context.
This process helps explain why qualia are difficult to isolate introspectively. The apparent simplicity of red conceals its continuous interaction with object structure, spatial organization, expectation, language, memory, and action. Even when an observer attempts to attend only to redness, the representation remains embedded in an active system that distinguishes it from other colors, locates it in a field, sustains the instruction to observe it, and compares the current state with preceding states.
Qualitative experience therefore has both a stable core and a changing surround. Iterative updating preserves enough of the core representation to maintain identity while allowing the associated context to evolve. This dynamic organization provides a mechanism for simultaneous stability and change within consciousness.
Implications for Artificial Consciousness
The theory has direct implications for evaluating artificial systems. Contemporary neural networks contain distributed representations, attention mechanisms, recurrent computation, context-sensitive activation, multimodal processing, and internal state. These features may approximate individual components of conscious architecture. Their presence alone does not establish phenomenal consciousness.
An artificial system satisfying the iterative phenomenalization criterion would require an integrated focus of attention whose contents persist through partial replacement. The retained contents would need to influence the selection of subsequent states. Internal representations would have to interact with multimodal sensory models, memory systems, goals, actions, and self-maintaining processes. The system would also require continuous endogenous activity rather than isolated episodes initiated only by external prompts.
The original model proposes that artificial working memory could maintain active internal representations, update them through multiassociative search, and combine them with sensory and motor systems in a global workspace. Such a system could potentially develop attentive continuity, progressive imagery, long-horizon reasoning, and a persistent model of its relationship to the environment.
Artificial qualia would depend on the representational organization of the machine. A system with visual processing modeled only as abstract labels might possess conceptual knowledge of color without anything analogous to visual color experience. A system containing recurrent, topographically organized visual representations integrated into an iterative workspace would be a stronger candidate. Its artificial redness would not need to reproduce human redness exactly. The quality would be determined by the architecture of its own perceptual representations and the manner in which they were incorporated into its continuing state.
Artificial superconsciousness could extend these properties beyond biological limits. A machine might maintain more representations simultaneously, preserve them across longer intervals, integrate more sensory modalities, and perform more iterative updates before losing contextual coherence. Its conscious field could have greater temporal depth, representational breadth, and associative reach than human consciousness. Such a system might experience composite or high-dimensional qualitative states that biological working memory cannot sustain.
Empirical Predictions
The iterative phenomenalization hypothesis produces several testable predictions. Conscious perception should correlate with the persistence of modality-specific representations across multiple partially overlapping neural states. Stimuli that are processed but remain unconscious should produce shorter, less recurrent, less globally integrated, or less state-spanning activity.
Perceptual masking should interrupt the incorporation of a representation into subsequent states. A masked color stimulus may generate early visual activity while failing to develop the sustained recurrent pattern associated with conscious chromatic experience. Increasing stimulus duration or attentional priority should increase the number of states across which the representation remains active.
Changes in perceptual vividness should correlate with the strength of recurrent interaction between sensory areas and higher-order working-memory networks. Vivid imagery should produce stronger or more stable sensory reactivation than weak imagery. Individuals with reduced visual imagery may maintain abstract items in working memory while showing diminished top-down reconstruction of sensory maps.
The hypothesis also predicts that perceptual fading should correspond to the gradual loss of state-spanning support. During Troxler fading or sensory adaptation, a stable visual representation may become less capable of recruiting itself into subsequent updates. Attention shifts, microsaccades, or contextual changes may restore the representation by increasing prediction error and renewed recurrent activation.
Neural recording experiments could examine whether consciously perceived features display longer representational half-lives and greater overlap between consecutive population states than physically similar unconscious features. Multivariate decoding could estimate whether the identity of a perceived feature remains recoverable across successive time windows and whether its persistence predicts subjective vividness.
Experiments could also manipulate the proportion of working-memory content retained across states. Increased retention should enhance perceptual continuity and depth while excessive turnover should produce fragmented or superficial conscious processing. These predictions connect the phenomenology of consciousness with measurable properties such as representational persistence, state overlap, recurrence, global connectivity, and context-sensitive influence.
Discussion
The iterative updating model offers a framework in which conscious experience emerges from the temporal organization of representational activity. Sensory information becomes conscious when it is incorporated into a succession of partially overlapping working-memory states. This process allows the representation to persist, interact with context, influence associative search, recruit memory, constrain perception, and guide future processing.
The model assigns distinct roles to representational differentiation and temporal integration. Modality-specific neural systems determine whether the content is red, painful, warm, melodic, or appetitive. Iterative overlap allows that content to occupy the experienced present. Phenomenal consciousness arises from their combination.
This approach provides a possible account of why a representation can guide behavior unconsciously in one condition and become experienced in another. The underlying feature-sensitive processing may occur in both cases. Consciousness depends on whether the representation becomes recurrently stabilized and carried forward as part of the active state of the organism.
The framework also preserves the intuition that consciousness is processual. Experiences cannot be fully identified with isolated snapshots because their apparent identity depends on continuity across time. A red percept occupies an interval during which chromatic information is maintained while its context evolves. The observer experiences a stable property because sufficient representational structure remains active across successive transformations.
The theory remains incomplete in several respects. It does not yet specify the precise spatial scale, temporal resolution, or degree of overlap required for consciousness. It must distinguish iterative phenomenalization from recurrent processes that occur unconsciously. It also requires a more detailed account of binding, self-modeling, affective valuation, and the relationship between global availability and local sensory recurrence.
These limitations define a research program rather than a terminal objection. The theory proposes measurable dimensions and architectural requirements. It can be evaluated by examining how representational persistence, partial updating, recurrent interaction, and global integration vary across conscious and unconscious conditions.
Conclusion
The experience of red need not be interpreted as an inexplicable property appended to a neural signal after visual processing is complete. Redness can be identified with the organization of a chromatic representation, while conscious presence can be identified with the representation’s sustained incorporation into an iteratively updated workspace.
On this account, subjective experience consists of representational activity carried through time. The contents of attention remain conscious by contributing to a sequence in which each state preserves part of the state before it and helps construct the state that follows. Iterative overlap creates a temporally extended field in which sensory, affective, mnemonic, and conceptual information can coexist, interact, and acquire phenomenal texture.
The central proposal can be summarized as follows:Phenomenal consciousness=representational differentiation×iterative temporal integration
Representational differentiation supplies the qualities of experience. Iterative temporal integration supplies their presence within a continuing subject. Together, they may provide a physical and computational account of how signals become sensations, how perceptual states become experiences, and how an artificial system could eventually enter a conscious stream of its own.
References
Baars, B. J. (1988). A cognitive theory of consciousness. Cambridge University Press.
Baars, B. J., & Franklin, S. (2003). How conscious experience and working memory interact. Trends in Cognitive Sciences, 7, 166–172.
Baddeley, A. D. (2012). Working memory: Theories, models, and controversies. Annual Review of Psychology, 63, 1–29.
Chalmers, D. J. (1995). Facing up to the problem of consciousness. Journal of Consciousness Studies, 2, 200–219.
Cowan, N. (1999). An embedded-processes model of working memory. In A. Miyake & P. Shah (Eds.), Models of working memory (pp. 62–101). Cambridge University Press.
Cowan, N. (2005). Working memory capacity. Psychology Press.
Dehaene, S. (2014). Consciousness and the brain. Viking.
D’Esposito, M., & Postle, B. R. (2015). The cognitive neuroscience of working memory. Annual Review of Psychology, 66, 115–142.
Edelman, G. M. (2004). Wider than the sky: The phenomenal gift of consciousness. Yale University Press.
Fuster, J. M. (2015). The prefrontal cortex (5th ed.). Academic Press.
James, W. (1890). The principles of psychology. Henry Holt.
Lamme, V. A. F. (2006). Towards a true neural stance on consciousness. Trends in Cognitive Sciences, 10, 494–501.
Reser, J. E. (2016). Incremental change in the set of coactive cortical assemblies enables mental continuity. Physiology & Behavior, 167, 222–237.
Reser, J. E. (2022–2024). A cognitive architecture for machine consciousness and artificial superintelligence: Thought is structured by the iterative updating of working memory. arXiv:2203.17255.
Rosenthal, D. M. (2004). Varieties of higher-order theory. In R. J. Gennaro (Ed.), Higher-order theories of consciousness (pp. 17–44). John Benjamins.
Seth, A. K. (2021). Being you: A new science of consciousness. Dutton.
Tononi, G. (2004). An information integration theory of consciousness. BMC Neuroscience, 5, 42.
Leave a Reply