The author proposes that phenomena attributed to attention could be explained more parsimoniously by appealing to factors such as priority signals and task sets (“biasing factors,” for short). Thus, no homuncular, central executive is needed to direct attention to mental representations. Instead, some representations are processed more deeply than others because of the biasing factors, which is consistent with the view of attention as an effect rather than as a cause (Krauzlis et al., Reference Krauzlis, Bollimunta, Arcizet and Wang2014). Thus, the inferred “observing agent” of the conscious field (the “perspectival self”; Seth, Reference Seth2021), which is a result of the geometry of the conscious field and cannot itself be observed in the field (Merker et al., Reference Merker, Williford and Rudrauf2021), might not be the actual agent(s) dictating which representations are processed deeply. To the “observing agent,” attention seems limited only because “the pattern of weighted inputs gives the appearance of limited sensory resources, because increasing the weight of one input necessarily involves reducing the proportional weight of all of the other inputs” (Krauzlis et al., Reference Krauzlis, Bollimunta, Arcizet and Wang2014, p. 458).

But is there a single observing agent? It has been proposed that the field is not apprehended (or “sampled”) by a single agent, but by a collection of (unconscious) “response codes,” each of which is activated differentially by the contents of the field (Morsella et al., Reference Morsella, Godwin, Jantz, Krieger and Gazzaley2016). (There is no homunculus fallacy here because each code, with its own concerns and action tendencies, does not replicate the function of the whole system.) It seems that there is a single observing agent, perhaps because of the way in which the field is formatted for each response code. We refer to this as the “Greek theater” effect. In Ancient Greece, theaters contained a central, circular stage. With this geometry, any spectator could believe that he or she had “the best seat in the house.” But the illusion occurred equally for every spectator. Similarly, the geometry of the conscious field, which is designed that way for the purpose of adaptive action selection (Morsella et al., Reference Morsella, Godwin, Jantz, Krieger and Gazzaley2016), would seem to each response code as if the field were directed only toward it. It is adaptive for things to appear as if from a first-person perspective. Whether there is a single conscious field observed by the response codes or each code possesses its own field is an interesting question (for some evidence for the latter, see Gazzaniga & LeDoux, Reference Gazzaniga and LeDoux1978; Rapp et al., Reference Rapp, Fisher-Baum and Miozzo2015).

The conscious field and the Greek theater effect would still arise if, for example, the verbal, “narrative self” were absent, as might occur during anesthetization of the left hemisphere in the Wada test. Moreover, “attentional” phenomena attributed to the self (e.g., sustained selective attention) would persist in the absence of, say, the “narrative self” (Seth, Reference Seth2021) or “left-hemisphere interpreter” (Gazzaniga & LeDoux, Reference Gazzaniga and LeDoux1978). A “perspectival self” and some representations being processed more deeply than others would persist.

The field is a “closed system” in the deepest sense. One infers that the allocation of attention is occurring because of processes located spatially “inside the head.” However, the head itself is a representation existing within the field (Köhler, Reference Köhler1947) along with, for example, percepts, urges, and “intra-psychic” conflicts (e.g., approach-avoidance conflicts). There is no direct conscious access to the mechanisms that generate conscious contents (e.g., percepts, urges, and memories), action conflicts, or that process some representations more than others. Conscious contents, which are usually “encapsulated” and cannot be turned off or modulated at will, wholly and exclusively determine “voluntary” action selection (Morsella et al., Reference Morsella, Godwin, Jantz, Krieger and Gazzaley2016). In voluntary action selection, there is no influence other than the contents of the field at that time. The combination of contents determines action selection. Thus, the field is unlike the “smart helmets” of fighter pilots. If one could monitor through telemetry what was projected onto a pilot’s shades, one would be unable to predict consistently what the pilot decides. For example, if the pilot felt dizzy and hence returned to base, the cause of the pilot’s decision would not be represented in the helmet. In contrast, voluntary action is not based on what occupies the conscious field plus information that exists in some other system, such as a system that could evaluate the degree to which, for example, the contents of the field are nonsensical. In the field, the absence of information does not imply information about absence, as is obvious with respect to visual scotomas, anosognosia, sensory neglect, and in the nonsensical dreams that we regularly accept as real and sensible.

The field reacts to the introduction of a stimulus, but it is the composition of the field as a whole, including memories, that determines action selection. At one moment, the field provides the “global best estimate” (Merker, Reference Merker, Shimon, Tomer and Neta2012) for this composition, and there is never any indecision for the action selection mechanism. Perhaps, conflicts reflect only thoroughness, a freebie from encapsulation. It might be that the observer infers that there is intra-psychic conflict because the contents activated by a stimulus are incongruent with each other. One might report, “I desired the cupcake because I enjoy cupcakes and skipped breakfast, but I was conflicted about eating it because of doctor’s orders” when it might be more accurate to state, “The cupcake activated in the my field thorough semantic knowledge that the stimulus was highly desired by me then (because I was hungry and like cupcakes) and my doctor forbids eating it.” These encapsulated, knowledge representations determine action selection just as other kinds of knowledge representations do. But the representations are not a window into the mechanisms giving rise to inclinations and behavior, much as the fuel gauge on one’s dashboard is not the actual tank of gas, and the perspectival self is not in charge of the allocation of attention.

Within this closed system, some contents are processed more deeply than others. As noted by the author, this is not due to a single homunculus directing attention toward one content versus another. Instead, this is due to biasing factors that operate outside of the conscious field.