The very existence of pure perceptual states may be transcribed as whether we can hear acoustic sensations or mere sounds or whether we can see pictorial sensations or mere images. Philosophically, this is usually put in terms of differences between seeing and seeing as; there is a sense of seeing (seeing pictorial sensations) independent of seeing as (seeing an apple). It is this former sense of seeing that presumably counts as pure perception. While seeing is theoretically neutral, seeing as is theoretically involved. Some theorists think that seeing as is inherent to seeing; there is no seeing without seeing as, and as seeing as involves conceptualization, there is a sense of seeing that is inherently conceptual (Fodor, 2007; Mandelbaum, 2018). This is, however, a questionable point. For example, in seeing an apple, the conceptual attribute apple is far from being a perceptible property but rather a property that is part of the subject´s conceptual repertoire⁵. But even without any conceptual repertoire, individuals can still perceive the world (e.g., infants and animals), so perceptual attributives should work independently of the conceptual repertoire of subjects. Here, I assume that seeing is inherently different from seeing as. If there is something like being in a pure perceptual state, such a state must be deprived of any conceptual attribute at all: it must be seeing and only seeing.

Everything points, then, to the fact that a pure perceptual state must be entirely composed of perceptual (and not conceptual) attributives —perception is, in these respects, non-conceptual⁶. The notion of nonconceptual content was initially coined (by Evans) as opposed to conceptual content, it is fundamentally contrastive and must be explained by distinguishing it from any particular notion of conceptual content. Evans (1982), for example, suggests that our conceptual abilities must be constrained by specific systematic abilities derived from the structure of thoughts. Evans called this the Generality Constraint. The idea is that “...if a subject can be credited with the thought that a is F, then he must have the conceptual resources for entertaining the thought that a is G, for every property of being G of which he has a conception” (Evans, 1982, p. 104). Although how to better characterize the notion of nonconceptual content remains inconclusive, I will take (following the contrastive understanding) that nonconceptual content must be such content that does not satisfy the Generality Constraint. Therefore, I assume here that when we get into a purely perceptual state, its content must be essentially non-conceptual. In other words, when we have a perceptual experience, its content must be alien to the concepts deployed by such an experience —when concepts appear, pure perception becomes impure.

Connected to the idea that the content of pure perception must be non-conceptual is the idea that its format must be iconic. The idea that pure perceptual states have an iconic format (while cognitive states have a discursive one) has been suggested by many theorists (see, for example, Dretske, 1981; Fodor, 2007; Carey, 2009; Burge, 2010, 2014; Block, 2014).

Consider, for example, the vehicle in which the picture of an elephant in the forest and the sentence “an elephant in the forest” are transmitted. The difference seems intuitively evident; the former is an image-like representation, and the latter a sentence-like representation. Following Quilty-Dunn (2020), there are two essential differences between iconic and discursive representations. Firstly, parts of icons correspond to parts of the picture (the part principle)⁷. And secondly, icons represent multiple features holistically (the holistic principle). By way of example, in a picture of an elephant in the forest, every part of the picture represents parts of the elephant (or the forest), and each part of the picture represents different properties (e.g., shape and colour). This is, however, not true for the sentence “an elephant in the forest” since no part of the sentence represents parts of an elephant in the forest, nor multiple features of the elephant or the forest⁸. The suitability of these principles is, however, far from being consensual. As occurs with non-conceptuality, how the notion of iconicity should be defined is not free of controversy. Burge (2018), for example, poses serious and disturbing problems to the arbitrariness in which representational units can be characterised. There is, in fact, no way to stipulate what corresponds to a representational unit. Nevertheless, I consider these explanatory principles useful, perhaps not for defining iconicity, but for contrasting iconic representations with discursive ones.

There is a reasonable way in which the content and the format of a pure perceptual state are mutually connected. Differences in content are, in fact, usually explained as differences in format. For example, when we claim that the type of content deployed in a pure perceptual state is nonconceptual all the way down, this claim includes that this content must be conveyed in an iconic format⁹.

Consequently, pure perceptual states must be entirely transmitted in iconic format. Understanding by icons those representations that do not represent-as, do not have a constitutive structure, logical form, or truth conditions. Icons only represent a likeness of the properties of analogically represented images. They are, in short, naturally related to the matter represented.

When we have a purely perceptual experience, there must also be something in such an experience that makes it phenomenologically peculiar and different from other experiences¹⁰. Indeed, pure perceptual states demand a distinctive phenomenology that differs from other mental states in its very composition. To the extent that the representation of a pure perceptual state is nonconceptual and iconic, its phenomenality must dovetail these components. Therefore, there should be a particular mental state whose phenomenal properties represent the perceived world in the way of an unconceptualized characterization of a perceptual experience. When biting into a lemon, seeing a sunset or hearing a piano, there is an enclosed particular component in these occurrences that makes perceivers meet in a trivial phenomenological position. This unconceptualized mental state must be necessary for the subsequent comprehension of the perceived object, and it must be phenomenologically distinct from the already conceptualized version of the object of perception.

Thus, in terms of ‘what it is like to feel perceptual objects’, there should be a substantial difference between the pure perceptual experience and its conceptualized version. For example, an astronomer conceptually grasps things through the telescope that a non-expert simply does not get, i.e., conceptual and discursive knowledge provides a specific phenomenology (Hanson, 1958; Kuhn, 1962; Vetter & Newen, 2014; Lupyan, 2015). However, the unconceptualized and iconic nature of pure perceptual experience must produce the same phenomenal experience both in experts and non-experts. Therefore, just as conceptual knowledge admits, in principle, unlimited phenomenological differences between subjects, the phenomenology of the un-intellectualized part furnishes —maintaining certain conditions fixed— an identical phenomenology for every individual (Fodor, 1983; Block, 2016). Crucially, the phenomenology of the conceptualized and discursive part must have the phenomenology of the unconceptualized and iconic part incrusted; i.e., the pure percept must be an essential part of the intellectualized one. This is not, in principle, true in the opposite direction since concepts and language cannot modify pure perception.¹¹

Therefore, any perceiver experiencing the same percept must, prior to the involvement of concepts and language, have the same phenomenal experience. Put differently, there is a class of nonconceptualized and iconic phenomenal properties such that holding the object properties and relations, the perceiving conditions (light, shadow, and so on) and the location of the focus of attention fixed, the phenomenal experience of two individuals (or the same individual at different times), are necessarily indistinguishable. There must be, in sum, an invariable and distinctive phenomenology in the space occupied by the nonconceptualized and purely iconic perceptual experience.

The core claim surrounding non-conceptuality is that if the content of PPS is nonconceptual, then pure perception ends when conception begins. At first glance, this is a significantly strong constraint since we can conceptualize objects incredibly fast. Research in perceptual categorization has found that subjects categorize objects as fast as 13ms from stimulus onset (Potter et al., 2014)¹². For some, these results suggest that some conceptual information might be part of the perceptual process (Mandelbaum, 2018); for others, the crucial point behind this evidence is that perceptual categorization must be achieved in a feedforward way since there is not enough time for feedback processing (Potter et al., 2014). Both interpretations seem to suggest that perceptual states comprise some kind of conceptual information, but other interpretations are possible. Indeed, for the nonconceptual characterization of PPS developed here, these results must be interpreted as if PPS's content is basically composed of the low-level properties of perceptual processing (i.e., shape, colour, size, texture, brightness, or motion), the higher-level properties (i.e., categorization, or identification), which are already conceptualized features, are not part of our purely perceptual characterization¹³.

All this fits perfectly with the most commonly accepted view on perceptual processing, which suggests hierarchical and sequential processing of perceptual properties: the visual system extracts increasingly complex visual features from sensory input to finally link it with conceptually stored representations (Felleman & Van Essen, 1991). But is the incredibly short time window until objects are recognized enough to capture all the low-level properties (shape, colour, size, texture, brightness, motion and so on) typically present in stimuli? Probably not. So, how many low-level properties are necessary to categorize objects? Perhaps just a few or even a single low-level feature might be enough for categorization. According to Fodor (1983, p. 97), for example, object categorization could be achieved via general shape properties. So, taking the non-conceptual part of the process as the only pure perceptual part, there will be cases where only one low-level property may be enough to reach the posterior identification and categorization. All this suggests at least two things: first, the hierarchical and sequential view should be reconsidered, and second, the pre-conceptual process can be incredibly exiguous.

Lastly, and more importantly, recent empirical data on the temporal dynamics of perceptual processing, both in behavioural (e.g., Kirchner & Thorpe, 2006; Crouzet et al., 2010) and neuroanatomical studies (e.g., Grill-Spector & Kanwisher, 2005; Groen et al., 2017; Harel et al., 2016; Ramkumar et al., 2016; Caddigan et al., 2017), indicates the co-occurrence of low and high-level representations. Grill-Spector and Kanwisher (2005), for example, found that subjects are sometimes equally quick at categorizing objects than merely detecting their presence; furthermore, others show that categorical information might be, on occasion, first encoded (Caddigan et al., 2017). So, if categorization and detection co-occur, there cannot be much space for the initial processing of low-level properties. So, conceptual representation not only does not require an exhaustive representation of low-level properties, but sometimes it is activated intermingled to, or even prior to such low-level properties. This co-occurrence is undoubtedly a challenging consequence for the characterization of PPS since rather than sequential and hierarchical processing that produces increasingly abstract representations, it suggests that the contributions of low and high-level information are, to some extent, inseparable. So, if conceptual information appears merged with the object´s structural information, it is hard to see how there can be a mental state uniquely composed of nonconceptual content.

Just as the content of a pure perceptual representation is restricted to nonconceptual content, its format is limited to iconic format. Recall that icons are subject to certain restrictions that make them peculiar; icons produce representations whose parts correspond to parts of the picture (the part principle) and represent multiple features holistically (the holistic principle). Let us see the case of face perception. When we see faces, there must be a mechanism that, before identifying a face belonging to (face recognition), represents the face as a face and not as another kind of thing (face detection)¹⁴. It must be, therefore, the configuration of the typical low-level attributes of a face that ultimately allows detecting a face —my stored “knowledge” that faces are made up of two eyes positioned above the nose that, in turn, is above the mouth makes me detect such percept as a face and not as another kind of thing. Thus, as face detection is already tainted with the “knowledge” that I am facing a face and not another thing, face detection should involve something more than a purely iconic representation¹⁵. This might suggest that representing faces and representing the typical low-level attributes of faces involve, at least partially, different mechanisms, and these low-level attributes are, presumably, the ones that are represented in a purely iconic fashion. So, just as with non-conceptuality, pure perceptual states must be, at most, restricted to the iconic representation of the low-level attributes.

Again, these restrictions substantially decrease the possibility of empirically isolating the iconic aspect of perception. One reason is that the processes previous to face detection are inaccessible to subjects’ consciousness in a way that face detection itself is not. After all, I can manage to be aware that what I am detecting are faces and not something else, but it is very hard, if not impossible, to be aware of the low-level properties that make a face a face and not something else. Furthermore, evidence shows that the low-level attributes (the only candidate for being a purely iconic mechanism) are processed holistically in conjunction with face detection. For example, some studies show that face detection is not affected by inversion effects (inverted faces) as much as face recognition (Lewis & Edmonds, 2003). This being the case, the alleged purely iconic aspect of face perception becomes ephemeral, undetectable and, most likely, imponderable and inseparable from the general mechanism of face detection. Thus, if we take face perception as triggered, at least initially, by holistic configurations (Bentin et al., 2006), then we might accept that the visual system cannot generate purely iconic outputs for detecting faces. In sum, as occurs with non-conceptuality, iconicity appears mostly merged with non-iconic aspects during perceptual processing.

According to PPS, there must also be something like being in a pure perceptual state that is felt singularly. I have already argued that having nonconceptual content and iconic format must constitute one of the main reasons for PPS's phenomenology being distinctive. It is, therefore, to be expected that the above problems raised for content and format will also extend to the particular phenomenology of PPS —if it is hard to capture pure perceptual representations with exclusively nonconceptual content and iconic format, then it will be equally hard to capture a particular way in which such representations are experienced. However, postulating a specific sort of phenomenology is a captivating way to grab the distinctive nature of PPS. Intuitively, it is different to see the brightness of a winter´s day, hear the soft tone of a familiar voice, or feel the silky touch of clean sheets than it is to plan a weekend vacation, solve math problems, or remember a phone number; there seems to be a solid distinction to be marked here. Disengaging the particular phenomenological aspect of PPS from the already conceptualized and discursive part is, in philosophical terms, to separate seeing from seeing as.

One way to assess this issue is by comparing the phenomenology of PPS with the phenomenology of other states, let's call them quasi-perceptual states. Introspective visualizations, episodic memories, mental imagery, dreams or hallucinations (all of which I will refer to as visualizations) are examples of non-perceptual processes whose phenomenology seems to be similar to the phenomenology of pure perceptual experiences. In a broad sense of perception, these mental states might be considered perceptual since they share remarkable similarities with paradigmatic perceptions: visualizations are in some way linked to sensory properties (like colour or shape), are experienced from the first-person perspective, are modality-individuated, share similar neural substrates than perception, engage the same pattern of eye movements, are at least partly represented in iconic format and nonconceptual content, and most likely are felt phenomenologically similar. However, unlike paradigmatic perceptions, visualizations are not triggered by corresponding sensory stimulation in the relevant sense modality (Kosslyn, 2005; Pearson et al., 2015; Nanay, 2015)¹⁶. This is a crucial difference: just as perceptual experiences require bottom-up processing, visualizations occur in a one-way top-down process¹⁷. It is intensely debated whether visualizations (specifically, mental imagery) belong to cognition, perception or is a sui generis psychological state; however, the critical point here is that, just as occurs in perceptual experiences, there should be a part of visualizations that is nonconceptual and iconic, and ultimately, phenomenologically similar to PPS's. Consider Perky's (1910) famous study, where participants look at various faintly projected objects in front of them just above the threshold of visibility while simultaneously visualising them mentally. The study shows confusion between the imagined and the real picture; subjects fail, in general, to distinguish between the contributions of the imagined and the perceived aspects of the experience¹⁸. This effect might, perhaps, be produced by failures in memory, but it also suggests that the phenomenal properties of perceived, imagined or dreamed experiences are similar. Studies of this sort provide evidence that the same kind of phenomenology present in PPSs might also be present, to some extent, in all sorts of visualizations. All this is clearly at odds with the idea that there are phenomenological differences between PPSs and other mental states.

In order to rely on introspection as a source of evidence that PPS constitutes a genuine mental state, we have to focus on the mental state we get into when for example, we observe an object or scene in a raw sense —i.e., observing without signification or seeing without seeing as. Try to observe any object around you by ignoring the concept to which it refers (by cancelling the reference or removing the conceptual attributive); it is a really complex task, perhaps even impossible. At least in our everyday perceptual lives, when looking from oneself, the phenomenal experience seems to have a conceptual label attached, as if it were not possible to observe any object beyond the concepts associated with that object. This is probably because, as we have seen above, conceptualization occurs quickly (almost instantaneously) and automatically —swiftness and automaticity are, perhaps, what produce the false impression that categories are indivisible parts of objects. But be that as it may, it is extraordinarily hard to individuate introspectively the mental state defined by PPS.

But perhaps one should be more rigorous. Perhaps one way to get into a mental state of this sort is by focusing all your attention on the essential properties of objects. For example, focusing attention on the brightness of a star in the sky or the intensity of the green colour of the forest should, in principle, count as entering into a nonconceptualized, iconic and phenomenologically peculiar perceptual state. But can we introspectively experience these properties independently of their referent or other contextual factors? Can we experience them in isolation? No matter how hard I try, I cannot see, from my first-person point of view, the brightness outside the star or the intensity of green outside the forest —the more I try to remove contextual information, the less I can. But even more, even if I were able to cancel much of such contextual information and manage to be in an apparently pure perceptual state (perhaps through some sort of meditative state), how can I be sure that I am introspectively experiencing this kind of mental state? How can I verify I am in a PPS if just at the moment I identify it ceases to be a PPS? And what is even more pressing, how can I show you that I have been in this state? How can I show another person that I have seen brightness and the intensity of greenish in its most elementary form? Getting rid of these concerns is, I think, not possible.

Against this, it can be argued that non-linguistic animals and pre-linguistic infants have perceptual experiences with contents similar to our own, but they lack the appropriate concepts to specify the contents of their experiences. Consequently, although infants and animals cannot introspect to observe their internal states, their phenomenal experiences must be similar to those supposedly defined by PPS. This is one of the most employed arguments in favour of nonconceptual content in perception (Peacocke, 2001). Against this, however, one can argue that infants and animals possess a rudimentary conceptual system that allows them to identify, recognize and discriminate objects. Some studies show that infants have selective attention to faces, that they can discriminate between faces from birth, and that they react to face inversion and other race effects (Otsuka, 2014). Furthermore, during the perinatal period fetuses can differentiate their mother voices from other voices (Carvalho et al., 2018) as well as recognize words (Gervain, 2018) or even perceive rhythm and synchronization (Provasi et al., 2014). And something similar occurs in the case of animals. Animals recognize, distinguish and identify things in the world; so, although they lack discursive representations in the sense of language-like elements, they can still possess rudimentary conceptual systems in order to identify and discriminate between objects.¹⁹ According to empirical research in animal cognition, this is a very reasonable position (for review see Seyfarth and Cheney, 2015).

In sum, separating PPS from the rest is, from the first-person perspective, beyond the scope of empirical scrutiny, so the introspective approach is doomed to fail.

Even admitting that from the first-person it is not feasible to individuate PPS, there are other perspectives to be considered. Indeed, the thesis of whether there should be something like being in a PPS has been widely evaluated from a psychological perspective (Hansen et al., 2006; Levin & Banaji, 2006; Balcetis & Dunning, 2013, 2010; Rolfs et al., 2013; Scholl & Gao, 2013; Firestone & Scholl, 2014; Valenti & Firestone, 2019). This perspective has, of course, its difficulties. The most pressing is relying on subjective perceptual reports. Experimenters cannot know what individuals perceive without asking them (or inferring) what they are perceiving. In this case, subjects´ perceptual experience is (or can be) contaminated by many different factors (judgements, memory or the focus of attention). Simply, just at the moment that subjects verbally infer the percept, the possibility of isolating it is lost —there is a fundamental gap between the perceived and the informed. So, if, as noted above, PPS cannot be discerned from the first-person perspective, then subjective reports, which are verbal expressions of the first-person experience, will suffer from the same problem. So, in principle, PPS seems psychologically unapproachable.

But there is more. When researchers have tried, in thoroughly controlled situations, to retain the mental state defined by PPS, they have found that there is no way of removing other non-perceptual factors. Indeed, there are a lot of recent experiments that, in their quest to isolate pure perceptual processes, have found that low-level perceptual properties such as colour, shape, size, contrast, brightness or motion are continuously influenced by high-level factors such as motivation, action, affect, emotion, previous beliefs, categorization or language²⁰. These experiments have been accused of a lack of rigour. In particular, they are accused of using an overly confirmatory research strategy, confusing pure perception and perceptual judgement, being contaminated by demand and response bias, being the result of mere changes in the low-level properties, being the result of peripheral attentional effects, or not sufficiently controlling the effects of memory and recognition (for details see Firestone and Scholl, 2016). However, note that once these constraints are applied, psychological experiments are orphaned of the necessary instruments to address their fundamental goal, i.e., psychological experiments cannot isolate pure perception because they are profoundly and inescapably constrained by methodological limitations (Masrour et al., 2015; Cermeño-Aínsa, 2020).

Aware of these methodological limitations, some researchers have tried to isolate pure perception by focusing on a different level of description. The fundamental point is that a pure perceptual state can be disclosed, not by appealing to intuitive distinctions or phenomenological differences but by appealing to functional divergences (Rolfs & Dambacher, 2016). Instead of dealing with the typical low-level properties of visual processing (colour or shape), this strategy deals with more elaborate and controversial aspects of perceptual experiences, such as the perception of causality (Rolfs et al., 2013) or the perception of animacy and intentionality (Scholl & Gao, 2013). Let me explore this suggestion in depth.

Rolfs and Dambacher (2016) claim to have achieved isolating perception in a causality detection task. The collision of two billiard balls or dominos falling one after another perfectly illustrates the causal connection between two events: a widespread phenomenon whose roots, whether perceptual or cognitive, have been in question for long (Michotte, 1963; Hume, 1967). In principle, this causality involves two components, one based on the stimulus (the spatiotemporal coincidence of the events) and the other based on inference (the continuity of action transferred from the first object to the second). Using a visual adaptation paradigm (a tool that identifies specialized neural populations for specific visual features), Rolfs et al. (2013) investigated whether the inferential component occurs on a perceptual level. Each experiment in this study is a step towards securing the conclusion. First, Rolfs et al. provide evidence that the representation of causation is subject to adaptation. After prolonged viewing of causal collision events, subjects judged events more often as noncausal, thus showing an adaptive effect in causality. Then, in the second experiment, they ruled out the possibility that other low-level visual features could explain the effect. And finally, they ruled out a cognitive explanation by showing that the effect is spatially localized in specific retinotopical coordinates. This is the crucial point: if the retina and visual cortex share a reference frame, then only stimuli that appear causal produce responses in these neurons. Now, under the premise that visual adaptation reduces the responsiveness of neural populations that encode primary visual features, researchers conclude that visual routines in the retinotopic cortex detect and adapt to these effects. This, according to Rolfs et al. (2013), results in the functional isolation of a perceptual process composed of two components and suggests that the continuity of action (typically considered a high-level process) falls into the realm of perception. Thus, according to these researchers, the functional approach confirms that there must be a specialized perceptual module for causality; there should be an innate cause detector.

The problem with this approach is that it rests on the claim that pure perception (pure vision, in this case), but no other mental states, is associated with retinotopic effects. It is, however, unclear that retinotopic effects are exclusively connected with pure visual functions. For example, the perception of faces also shows adaptation and retinotopic effects (Webster & MacLeod, 2011) and, as seen above, it cannot be considered purely perceptual. In addition, some studies also reveal that retinotopic information interacts with category selectivity (Uyar et al., 2016), a process mediated by concepts. These studies suggest the presence of functional connectivity between retinotopic cortical areas and other brain areas that, though linked to perception, are not purely perceptual. Therefore, one can resort to the functional level to connect certain mental states with specific perceptual purposes, but this is far from being the isolation of pure perceptual processes. There are many open spaces in this research topic, but the alignment of adaptation and retinotopic effects with pure perceptual processes seems inadequate.

Finally, another way to individualize PPS is by grouping different lines of evidence; evidence taken separately is not persuasive, but taken as a whole, takes on much more strength. Scholl and Gao (2013) appeal to the compendium of reasons (five reasons) to argue that the perception of animacy and intentionality reflects concrete signatures associated exclusively with perception. However, all these reasons can be contested. Let’s see them one by one:

(1) The perception of animacy and intentionality is phenomenologically similar to other typical perceptual processes (e.g., depth or colour): it is also perceived effortlessly and automatically. Against this, we must remember that perceptual recognition is also achieved effortlessly and automatically, and yet it is not purely perceptual. Furthermore, as we have already discussed (and Scholl and Gao admit), introspective intuitions are poor guides on how the mind actually works. So, here we have a small (very small) step towards the idea that by isolating the perception of animacy, we are individuating a pure perceptual process.

(2) Some forms of perceived animacy and intentionality are subject to certain subtleties more representative of pure perceptual states than higher-level judgments. For example, chase detection is driven by a strict reliance on subtle visual details: like other typical perceptual properties, chase detection is subtly controlled by visual input. However, this is not very convincing. For example, reading also involves a very subtle pattern of visual details and largely relies on visual input, but no one would claim that reading is a purely perceptual process.

(3) Perceived animacy and intentionality are subject to implicit influences on visual performance. For example, in the perceived chasing case, subjects are not deciding which features should matter for detecting animacy; they merely perceive it. There seems to be an implicit underlying ability to detect animacy. But again, this is not a sound way to isolate pure perception. In the same way, one does not decide to identify objects or have obsessive thoughts about past events, which are also automatic and implicit processes that are far from being perceptual. Being automatic, implicit or even mandatory is by no means sufficient to be considered purely perceptual.

(4) Perceived animacy and intentionality result from the activation of visual brain areas. This is not a plausible way to isolate perception for at least two reasons: first, because the neural correlates of perceived animacy (just as Scholl and Gao acknowledge) remain largely unknown, and second, because physiological studies have widely shown the existence of top-down effects at very early stages of sensory processing (the following section will analyse this strategy).

(5) Perceived animacy and intentionality interact with other visual processes. Researchers focus on selective attention, a primary form of attention in online visual processing. Shifts of attention in perceived animacy (they continue with the example of chasing) are, just as in typical visual processing, very fast and involuntary. But again, perceptual recognition is also very fast and involuntary, and yet cannot be appropriately labelled as pure perception.

Thus, the same compendium of reasons drawn up to show the pure perceptual nature of animacy and intentionality can also be employed to argue for their non-pure perceptual nature. We can start from a pre-established conception of what pure perception should be, what properties it should be composed of, or what specific signatures should be constituted, but upon closer inspection, we realize that we are trying to isolate a mental state whose manifestations only make sense when joined with other non-perceptual mental states. In short, Scholl and Gao brilliantly demonstrate that the perception of animacy and intentionality deploys many signatures typically associated with visual processing, but they have not cleanly offered a demonstration of how pure perceptual mental processes can be individuated —their five lines of evidence can be contested.

In sum, psychological scrutiny is not a good indicator of the presence of a pure perceptual state. The above discussion suggests that psychological evidence cannot yield solid conclusions because there might be a fundamental dissonance between what an experimenter observes and what the subject is actually experiencing. Grasping the representational structure of PPS might be, in short, beyond psychological scrutiny.

The last way to elucidate the mental state represented in PPS is by looking into the brain. The hypothesis is that there should be a specific pattern of neural activation exclusive to PPS, whose stimulation is independent of the stimulation generated in other brain parts. In other words, there should be a way to isolate the brain activation elicited by PPS from the activation of the rest of the brain. At first glance, this is a problematic hypothesis since even accepting that specific patterns of neural activation (the PPS patterns) work alien to the influence of other brain parts, it is unlikely that these patterns operate alone. This is because the system’s operativity requires the simultaneous activation of many different brain areas. Nevertheless, the only real requirement for claiming that PPS has been neurologically isolated is that the brain areas responsible for the properties postulated for the PPS operate without the direct influence of other brain parts. Put differently, during the activation of the areas implicated in PPS there can be no top-down projections from higher brain areas; the information must exclusively flow in a bottom-up manner.

Note that this point connects with the possibility outlined in footnote 17 about including in our characterization of PPS the requirement that the information must exclusively propagate in a bottom-up manner. Recall that one way to get rid of annoying cases, such as visualizations (usually taken by psychologists as perceptions without direct external stimulation), is the inclusion of the bottom-up clause. So, to be in a PPS is, therefore, to get into a mental state whose content is nonconceptual, iconically conveyed, phenomenologically peculiar and whose information is transmitted exclusively in a bottom-up manner. The most recent neuroscientific research, however, disproves this possibility. When cognitive neuroscientists look into the brain, they observe that, except for the retina, all the visual brain is affected by top-down feedback projections not only from higher-visual areas but also from nonvisual areas (Gilbert & Li, 2013; Rolfs & Dambacher, 2016; Bar & Bubic, 2013; Ellis, 2019; Veniero et al., 2021; Jastrzębowska et al., 2021). This suggests that no part of the visual brain is conceived uniquely to represent PPSs. It is estimated that each brain area is connected to 66% of the rest of the brain (Markov et al., 2013), and only 10% of the synapses that reach the primary visual cortex originate in the thalamus, which brings sensory information from the retina, the remaining 90% are originated in the cortex itself (Peters, 2002, p. 184). Therefore, we could concede that the information incoming from the higher levels strongly influences the processing of the early cortical levels (for detailed examples of these feedback projections, see Cermeño-Aínsa, 2020, p. 9)²¹. Since the properties assigned to PPS are expected to be encoded during the early stages of perceptual processing, and the higher levels are continuously influencing the information processed during these early stages, the isolation of the processes carried out in PPS becomes really hard. Therefore, the isolation of pure perceptual states, even from the cerebral perspective, is unlikely.