Against the “ non-sensory ” view of affective valence Contra a visão “ não sensorial ” da valência afetiva

Valence is a key construct in the affective sciences and in the philosophy of emotion. Carruthers (2011, 2017) has recently offered an account of the nature of valence. He defends a (representational) version of what might be called the non-sensory signal theory of valence (NSS). According to the latter, valence is identified with inner signals—which are not themselves perceptual nor conceptual states of any sort—which mark sensory representations as good or bad. In this paper, I argue that Carruthers’s version of NSS is problematic on its own, independently of the plausibility of competing theories of valence. Carruthers’s arguments to the effect that valence is non-sensory fail to rule out the hypothesis that, together with arousal, valence might also be grounded in bodily, sensory representations. Carruthers’s claim that valence is not a sensory item in the furniture of the mind needs to be then more thoroughly substantiated.


Introduction
Emotions are classified as either positive emotions or negative emotions in virtue of the chara er of their valence.Not only emotions have valence as a component.For example, "homeostatic motivations" also exhibit a positive and negative chara er, such as hunger and thirst, which are negatively valenced.Moods are also valenced, such as depression and anxiety, which typica ly have negative valence as a component.In fact, valence (together with arousal) is a defining dimension of affective states in general.As Car uthers (2017) emphasizes, determining the nature of valence is key then for understanding the nature of affect.
Car uthers (2011,2017) has recently offered an account of the nature of valence.He defends a (representational) version of what might be ca led the non-sensory signal theory of alence (NSS).According to the latter, valence is identified with inner signals-which are not themselves perceptual nor conceptual states of any sort-which mark sensory representations as good or bad.In line with a certain tradition in the affective sciences, in this kind of account valence is then taken to be something that "attaches" to sensory/perceptual representations, so valence is something "extra" to sensory representations themselves.
In this paper, I ar ue that Car uthers's version of NSS is pro lematic on its own, independently of the plausibility of competing theories of valence.Car uthers's ar uments to the effect that valence is non-sensory fail to rule out the hypothesis that, together with arousal, valence might also be grounded in bodily, sensory representations (i.e.interoceptive representations).Car uthers's claim that valence is not a sensory item in the furniture of the mind needs to be then more thoroughly substantiated.Thus, even though NSS might be the case, Car uthers ar uments are not compe ling in showing why such claim might be on track.
I begin by briefly chara erizing the notion of valence.In the next section, I present the relevant a ects of Car uthers's version of NSS.That is, I thoroughly present and motivate only those a ects of Car uthers's view which are the target of my ar uments: the claim that valence is a non-sensory signal.Fina ly, I discuss why Car uthers's ar uments for his version of NSS fail to show that valence is a non-sensory phenomenon.

Characterizing valence
We strive to have certain kinds of emotions and we strive to avoid having other kinds of emotions.Certain emotions are agreea le, while other emotions are disagreea le.Certain emotions feel good, while other emotions feel bad.That is, there are positive emotions and negative emotions.For example, joy, pride, love, and amusement typica ly are positive emotions, while anger, fear, uilt, and contempt typica ly are negative emotions.Emotions are classified in this way in virtue of the chara er of their valence.Certain emotions are positive emotions since they have as a component positive valence, and certain emotions are negative emotions since they have as a component negative valence2 (e.g., Barrett, 2006;Prinz, 2004Prinz, , 2010)).
Valence is not only part of our fo k psychological understanding of the nature of emotion, but it is also a construct that plays a fundamental role in the scientific study of emotion (see, e.g., Bar ett, 2006;Russe l, 2003;Ber idge and Kringe bach, 2015), to the point that, for some theorists, valence is one of the main building locks of emotion and affect in general (Bar ett, 2006;Russe l, 2003).So note that the notion of valence in which I am intere ed in this paper is a non-normative notion that plays an explanatory role in psychology.Thus, contrary to a possi le reading of what some researchers have pointed out (e.g., Charland, 2005;Picard, 1997;Solomon, 2003), when it is said, in the affective sciences, that an emotion is positi e or negati e (i.e. that it has positive or negative valence) it is not being said that such an emotion is positive or negative in the sense of being goo or ba normatively, in any ethical or prudential sense.In the sense in which I am intere ed in this paper, valence is neither an ethical nor a prudential construct; it is a descriptive psychological construct that plays an explanatory role in the affective sciences.Then, just as Car uthers (2017) does, I am going to simply assume that there is such a thing as valence and that it does play a role in our best cur ent theories about emotion and affect in general.

The non-sensory signal theory of valence: Carruthers's version
What might be ca led the non-sensory signal theory (NSS) (e.g., Car uthers, 2011Car uthers, , 2017;;Prinz, 2004Prinz, , 2010) ) identifies valence with motivating inner signals which mark sensory representations as good or bad (wanted or unwanted).The signals in question are not themselves perceptual nor conceptual states of any sort.
These views align with a certain tradition in the affective sciences which consists in regarding the affective, valenced a ect of sensory/perceptual experiences as something that "attaches" to sensory/perceptual representations.In other words, the valenced a ect of a sensory experience is regarded as something "extra" to the sensory representations themselves.For example, according to the tradition in question, eating a sweet cake feels good because an affective mental item (valence) gets attached to the sensory representation of sweetness, where the latter being a distinct mental item from the former.That is, only when sensory/perceptual representations (e.g., sweetness, a landscape, music, etc.) have a "hedonic gloss" a ded by affect is that those representations become something that feels good (or bad).Such a "hedonic gloss" is considered to be a non-sensory item in the furniture of the mind, distinct from any sort of sensory/perceptual representation or high-level piece of knowledge (see, e.g., Berridge and Kringe bach, 2010, p. 9).
In this paper, I focus on critica ly discussing Car uthers' s version of NSS for the fo lowing reasons.Firstly, Car uthers' s version of NSS is the most recent philosophical proposal on the nature of valence.Secon ly, NSS has several explanatory advantages (Prinz, 2004(Prinz, , 2010)), and ar ua ly for this kind of theory the philosophica ly most careful ar uments have been developed.It is relevant then to show that Car uthers' s ar uments to the effect that valence is non-sensory are pro lematic, so that the version of NSS in question turns out not to be fu ly compe ling as it stands.Now, as I mentioned above, NSS aligns with the traditional view that valence is something that "attaches" to sensory/perceptual representations, endowing the latter with a "hedonic gloss" that makes such representations feel good or bad.Then, by ar uing that Car uthers' s version of NSS needs to be substantiated, I wi l be also su gesting that the relevant a ect of the tradition in question should be revised.
Let's get down to business.According to Car uthers (2011, p. 126-135; 2017), valence consists in an inner non-sensory signal that confers value (good or bad) to attended stimuli.This non-sensory signal inherently motivates the pursuit or avoidance of such stimuli.Even more, in Carruthers's account, it is assumed that valence is (nomologica ly) essential for motivating a l sorts of intentional a ions in the external environment.Consequently, in Car uthers's view, valence signals get genera ly attached to representations of external events (e.g., your partner ar iving home safe), rather than to inner bodily states, as in other similar views on the nature of valence (e.g., Prinz, 2004Prinz, , 2010)).This makes that the link between valence and intentional a ion in the environment is much closer compared to other accounts (see Prinz, 2004, p. 191-196).
More importantly, Car uthers (2011, 2017) a heres to the genera ly accepted view that affect consists in valence and arousal (e.g., Bar ett, 2006;Bar ett and Bliss-Moreau, 2009;Russe l, 2003) and that emotions have affect as a component.He also a heres to the view that affect is to a major extent dependent on inner bodily perception (i.e.interoception).Nonetheless, as I mentioned above, in Car uthers's view, valence signals are non-sensory signals.For, according to Carruthers, valence and arousal are separate causal mechanisms in the furniture of the affective mind and represented physiological changes constitute arousal, not valence.When we experience physiological changes, we do not then have the experience of valence, but of arousal.However, valence makes attended events good or bad, making thus a contribution to the phenomenology of the experience of such events.
Furthermore, in Car uthers's view, valence is also a non-conce tual signal.For it confers value without deploying high-level abstract knowledge, such as the concepts GOOD or BAD.In this account, the positive and negative valence value of a certain experience is a non-conceptual representation of the goodness and badness of the object of such an experience, re ectively.In a word, in this view valence is a non-sensory, non-conceptual indicator of value.
Car uthers emphasizes the role that valence plays in decision-making.Closely fo lowing Damasio (1994), Car uthers claims that, during decision-making, valence signals get attached to representations of considered options, making the latter attra ive or repe lent.In Car uthers's account, during decision-making, valence signals serve as a "common cur ency" that a lows the system to compare the value of options and choose among them.That is why patients with lesions in regions which Car uthers associates with valence, such as the orbitofrontal cortex (OFC), show poor decision-making capacities (Damasio, 1994).However, contrary to Damasio (1994), Car uthers holds that valence does not amount to sensory, interoceptive representations.
Car uthers (2011,2017) offers no positi e evidence for the view that valence is non-sensory.However, Car uthers (2011) does offer positive evidence and ar uments for the claim that valence is non-conceptual.He offers as evidence the famous Iowa Gam ling Task studies.As Car uthers remarks, in these studies subjects see some decks as bad, without conceptua ly judging them to be bad.Thus, so the ar ument goes, valence is not conceptual in nature.Among other aruments for the claim that valence is non-conceptual, Carruthers (2017) ar ues for this claim based on the assumption that phenomena ly conscious states only have non-conceptual contents and that valence can be phenomena ly conscious.Thus, so the ar ument goes, valence must be non-conceptual.However, considering that I have no quar els with the claim that valence is non-conceptual (and representational) and I agree with it, I am not going to discuss what this piece of evidence su gests regarding the nature of valence and the soundness of this ar ument.In this paper, I am concerned with the claim that valence is non-sensory.
Even though Car uthers offers no positive evidence for the view that valence is non-sensory, he does have ar uments against the view that valence is sensory.I discuss these ar uments below.

Problems for Carruthers's view
Car uthers's ar uments to the effect that valence does not amount to a sensory signal are pro lematic.As I commented above, Car uthers (2011,2017) endorses the genera ly accepted view that affect-i.e., "core affect" (e.g., Bar ett, 2006)-is constituted by valence and arousal.Now, according to Car uthers, valence and arousal are separate causal mechanisms in the furniture of the affective mind.While represented physiological changes ("interoceptive percepts") constitute arousal, valence amounts to a non-conceptual indicator of value, not grounded in any sensory modality.Let's ca l this the inde endence claim: valence and arousal are separate causal mechanisms of affect.
One of the main reasons Car uthers puts forward to support the independence claim is that dimensional ap roaches to emotion ar ange emotions in a circumplex graph, in which the "core affective states" of valence and arousal are represented as independent, orthogonal dimensions (e.g., Bar ett, 2006;Russe l, 2003).Thus, "it is implausi le that the former should reduce to the latter" (Car uthers, 2011, p. 130).Considering that arousal is the dimension of "core affect" which is uncontroversia ly grounded in sensory, interoceptive representations, valence should then be taken to be a non-sensory signal, not grounded in interoception.
A key pro lem with this ar ument is that it mistakenly takes reports of affective states to evince the nature of the causal mechanism responsi le for affect.Proba ly because, in some dimensional ap roaches to emotion (e.g., Russe l, 2003;Bar ett, 2006), lin uistic analyses of questionnaires and extended re orts of valence and arousal are graphica ly represented as orthogonal desc ipti e dimensions of affect in a circumplex graph (e.g., Russe l, 2003), the above ar ument mistakenly takes valence and arousal to be independent causal components of "core affect" (Bar ett, 2006), realized by separate causal mechanisms.That is, it is simply a mistake to infer that valence and arousal are realized by independent causal mechanisms of "core affect" from the fact that lin uistic analyses of reports of "core affect" result in a circumplex graph, where valence and arousal fi ure as orthogonal descriptive dimensions.The circumplex model is designed to capture what people say about their own psychology, about their own subjective experience regarding affective states.The circumplex model does not then track the way in which the causal mechanisms responsi le for those reports are related.Dimensional theorists of emotion explicitly recognize this point (e.g., Kup ens et al., 2013;Russe l and Bar ett, 1999).In fact, nothing in the circumplex model of affect, which Car uthers (2011,2017) endorses, prevents that valence and arousal are essentia ly unified at the causal level, both being grounded in bodily, interoceptive representations.In other words, nothing prevents that, for example, arousal is a component part of the mental state that constitutes valence, or that "core affect" amounts to valence at certain levels of arousal, i.e. that, at the causal level, the latter is not an independent a ect of "core affect" , but simply amounts to the intensity that valence can take.Note that I am not endorsing the view that valence is grounded in interoceptive representations.I am only claiming that the ar ument in question fails to rule out this hypothesis.
Certainly, as some dimensional theorists have found (Bar ett et al., 2004), one can consciously focus more on one descriptive dimension than the other, as Car uthers (2011) notes.However, let me insist, nothing in the circumplex model of affect prevents both valence and arousal from being realized by represented physiological changes ("interoceptive percepts") at the causal level.In fact, this is what dimensional theorists seem to endorse (e.g., Russe l and Bar ett, 1999, p. 814-815;Bar ett, 2015, p. 45;2006).Car uthers fails to consider then the possibility that valence and arousal might be essentia ly unified.As Bar ett and Bliss-Moreau remark, Core affect is a state of pleasure or displeasure with some degree of arousal (Barrett, 2006;Russell, 2003;Russell and Barrett, 1999).Together, valence and arousal form a unified state, so although it is possible to focus on one property or the other, people cannot feel pleasant or unpleasant in a way that is isolated from their degree of arousal (Barrett and Bliss-Moreau, 2009, p. 171).
For example, it could be the case that valence and arousal are unified in the sense that arousal, instead of being a separate construct from valence, simply cor esponds to the intensity or "volume" taken by the perceived physiological changes which (hypothetica ly) realize valence, so that arousal cannot take place without valence being constituted.
The fact that one can consciously focus more on one descriptive dimension (arousal) than the other (valence) wi l not do the job for the defender of the independence claim.Car uthers is impressed by this fact: people who are better at detecting their own heartbeats tend to exhibit more arousal focus than alence focus (Bar ett et al., 2004).The notions of arousal focus and valence focus simply refer to the emphasis that subjects place on words related to arousal and valence during emotion reports, so that arousal and valence emerge as important a ects in the verbal descriptions of affect in an individual over time (Bar ett and Bliss-Moreau, 2009).Then, considering that heartbeat detection is taken to be an indicator of interoce ti e accuracy (see Garfinkel et al., 2015), one might be led to conclude that arousal, rather than valence, is the a ect of "core affect" that is grounded in interoceptive perception (see also Dunn et al., 2010).
The above ar ument does su gest that the claim that arousal is closely related to bodily perception might be the case.However, notice that I am not disputing this claim (in fact, I agree with it).I am concerned here with the nature of valence, independently of the issue of the nature of arousal, and the above ar ument only su gests that arousal is closely related to bodily perception.More precisely, the above ar ument does not eak against the claim that valence could also consist in bodily perception.The above evidence is consistent with the claim that, for example, valence consists in the perception of a pattern of inner bodily changes, which includes several changes besides changes in heart-rate, so that a single varia le (e.g., heart rate) is valence-neutral by itself.In other words, the evidence in question is consistent with the hypothesis that valence value is determined by the overa l wholebody shape taken by the evolving inner bodily landscape, in which several physiological dimensions interact (e.g., Damasio, 1994, p. 263;2003), while arousal is determined only by heart rate perception (or the perception of heart rate plus a sma l subset of physiological changes).If this were the case, valence would also be grounded in a sensory system, contrary to NSS.Thus, this kind of evidence wi l not do the job for the defender of NSS.Certainly, heart rate seems to be a relia le indicator of arousal.However, heart-rate is just one dimension of a pattern of changes in the inner physiological milieu.Thus, considering that changes in heart rate might be critical for arousal but not much for valence value, it is certainly expected that people who are good at paying effortful attention to their heartbeats also exhibit an emphasis on the descriptive dimension of arousal during verbal reports.But this fact, let me insist, is silent with re ect to whether valence is non-sensory at the causal level.
It could sti l be ar ued that the claim that valence consists in the perception of such patterns of inner bodily changes rests on a confusion, because arousal, but not valence, is the a ect of affect that is traditiona ly taken to consist in such a pattern of bodily changes.Car uthers endorses a view along these lines: arousal is constitutive of the "fight or flight" preparations undertaken by the body in response to threat.[…] It consists of a variety of autonomic changes in heart rate, blood pressure, activity in the sweat glands, and levels of adrenaline and other chemicals in the bloodstream, as well as behavioral changes in posture, muscle tension, breathing rate, and so on (Carruthers, 2011, p. 127).
Note that this sort of view is precisely what "arousal theory" proposed under the label "general sympathetic arousal" .However, even though this conception of the notion of "arousal" is sti l rather uncritica ly endorsed in some corners of psychology, this conception of arousal is not tena le anymore.Let me briefly explain.
According to arousal theory, general sympathetic arousal underlies such "fight-or-flight" responses via a single mechanism that controls several measures of sympathetic/ autonomic effectors.This conception of arousal emerged from mid-twenty century research on the brainstem reticular formation, which was hypothesized to be realized by the soca led ascending reticular a ivating system, basis of the sort of a ivation (arousal) responsi le for the "fight-or-flight" responses that Car uthers has in mind.It was thought that the brain structure in question was a functiona ly homogenous structure and that it had a ivational (arousal) effects without any sort of ecificity.
This conception of arousal is deeply pro lematic.A key prediction of this ap roach is that physiological measures of sympathetic a ivity (such as electro-dermal response and heart rate) should significantly co-vary within and across individuals.However, that turns out not to be the case (see Berntson and Caciop o, 2007;Caciop o et al., 1991;Lacey, 1956Lacey, , 1967)).On the other hand, the many varia les involved during autonomic control do not exhibit some sort of single continuum of a ivation or arousal that could be involved in a simple "fight or flight" mechanism (Berntson and Caciop o, 2007).In other words, there is no patterned set of autonomic responses that constitutes a unified arousal system.In fact, the reticular formation, the sup ose ly key functiona ly homogeneous neural basis of the arousal system, is composed of several structures, each of them with its own functional profile (Sarter et al., 2003).Thus, the notion of arousal on which Car uthers relies seems to be pro lematic.In fact, it has been shown that certain autonomic measures do not reflect arousal at a l, but rather reflect valence properties.For example, cardiac a ivity, lood pressure, and skin conductance duration are likely to reflect affective valence (see Caciop o et al., 2000) (also the startle response and facial EMG indicate valence rather than arousal; Mauss and Robinson, 2009).In a word, considering that the notion of arousal as general sympathetic a ivation does not hold, the claim that arousal, but not valence, consists in the perception of a pattern of physiological changes is undermined.
Car uthers's view faces another related pro lem.Remember that one of the reasons that Car uthers puts forward for the independence claim-i.e., the claim that valence and arousal are separate causal mechanisms, being arousal the one constituted by represented physiological changes-is that interoceptively accurate people (Garfinkel et al., 2015) tend to exhibit more arousal focus than valence focus (Bar ett et al., 2004).
As I mentioned above, interoce ti e accuracy is typically measured by heartbeat detection tasks.In one version of this task, subjects are asked to determine whether their own heartbeats are synchronized with a metronome (Bar ett et al., 2004).Interoceptive accuracy is also measured by asking subjects to count their heartbeats, and then their responses are compared to the actual number of heartbeats as measured by ECG (Ehlers and Breuer, 1992;Schandry, 1981).Interoceptive accuracy only te ls us then how good an individual is in effortfu ly attending and keeping track of the consciously accessi le outputs of interoceptive processing (i.e.already formed "interoceptive percepts").In this sense, this kind of task is meta-representational: subjects must form "beliefs" about already formed interoceptive representations.Thus, interoceptive accuracy te ls us nothing about the causal mechanism of interoceptive percept formation itself, or whether it is working properly (i.e.delivering proper interoceptive percepts), neither whether it is hyperfunctioning or hypofunctioning in individuals high in arousal focus.These a ects are the relevant ones, if the goal is to determine whether either arousal or valence (or both) are constituted, at the causal level, by the perception of the physiological inner milieu.Then, the fact that individuals high in arousal focus tend to exhibit high interoceptive accuracy is silent about whether either arousal or valence (or both) are constituted, at the causal level, by bodily perception.In order to conclude that arousal, but not valence, is constituted by interoceptive perception, it needs to be shown, at least, that interventions in the functionings of the causal mechanism responsi le for interoception give rise to modifications in arousal, but not in valence, and that tri gering a state of arousal determines modifications in the interoceptive system, without altering valence properties.Considering what the constructs of interoceptive accuracy and arousal focus rea ly te l us, they are not much useful for the defender of the independence claim.It is worth also considering that, insofar as it is operationalized by heartbeat detection tasks, interoceptive accuracy does not reflect what might be ca led "general interoceptive accuracy" .That is, accuracy not only with re ect to heartbeat a ivity, but with re ect to the a ivity of the whole pattern of physiological varia les that constitute the physiological landscape of an organism.A more interesting cor elation then for the defender of the independence claim-even though not much useful for the reasons presented in the above paragraph-would be a cor elation between "general interoceptive accuracy" and arousal focus.This is the case since, as I mentioned above, the claim that arousal is determined only by heart rate perception is compati le with the claim that valence is grounded in interoceptive perception.The idea here is that heart rate perception is critical for arousal, while valence is constituted by the perception of a whole pattern of physiological changes, and not just by the single varia le of heart rate.Thus, the fact that good heartbeat detectors exhibit high arousal focus does not point towards the independence claim.The defender of the latter would prefer to find a cor elation between good whole-body perceivers and arousal focus.There is no such evidence.Moreover, as it is usua ly remarked, it is unclear whether consciously monitoring heartbeats indicates interoce ti e accuracy or a somatic, exteroceptive capacity: The "beats" that are monitored by subjects during the heartbeat detection task could simply be reflecting the a ivation of somatic, non-interoceptive receptors on the chest wa l.Ap ealing to the cor elation between arousal focus and interoceptive accuracy wi l not do the job for the defender of the independence claim.
Car uthers (2011, p. 130) puts forward another ar ument for the inde endence claim.As is we l-known, after the lesion, OFC/vmPFC patients lose their capacity to respond ap ropriately to rewards and punishers, which has severe consequences for their personal lives and social intera ions (Damasio, 2003).However, they retain their ability for cold reasoning.The standard explanation of the behavior of OFC/ vmPFC patients is that they lose the capacity to associate valenced responses with representations of behavioral options, which is key for normal decision-making.The upshot of this is that valenced responses are required for decision-making, and that malfunctioning of OFC/vmPFC compromises the ability to use such responses to uide decision-making.Carruthers ar ues that if valence were grounded in representations of bodily changes, people who tend not to explicitly focus attention on their heart rate would show aber ant decision-making abilities in the way shown by patients with OFC/ vmPFC lesions.But they do not.Then, so the ar ument goes, valence is not grounded in representations of bodily changes.
Valence should then amount to a non-sensory signal, while arousal is the dimension of "core affect" that is grounded in representations of bodily, physiological changes.
There are several pro lems with this ar ument.Let me point to two of the most salient of them.In the first place, the fact that OFC/vmPFC patients show poor decision-making abilities seems to favor the view that valence is grounded in representations of the body.Determining that something is positive or negative, beneficial or harmful, is key for decision-making.In Damasio's account, given its rich connections with regions involved in visceral representation and control, the OFC/vmPFC, during decision-making, plays the role of linking representations of external situations with representations of bodily, physiological responses.In this account, bodily responses are precisely the kind of responses that inform about whether something is beneficial or harmful: positively valenced bodily responses assign positive value to considered behavioral options, while negatively valenced bodily responses assign negative value to them.Insofar as OFC/vmPFC patients fail to link the valenced input from the body that informs about value with their considered options, they fail to behave ap ropriately.That is, the standard explanation of the pattern of behavior shown by these patients assumes that bodily responses determine valence value (Damasio, 1994), rather than determining arousal or another affective construct.
In the second place, it is simply a mistake to infer that a certain mechanism is malfunctioning from the fact that its outputs tend not to be explicitly attended.Thus, just as the fact that some people do not tend to explicitly focus on the phonetic properties of their spoken lan uage does not imply that their mechanism of lan uage production is malfunctioning, the fact that some people fail to explicitly focus on their heart rate does not imply that their mechanism responsi le for bodily representation is malfunctioning.It is not surprising then that such people do not behave as OFC/vmPFC patients.
Fina ly, fo lowing Schroeder (2004), Car uthers (2011) also ar ues that valence value is not determined by interoceptive representations on the basis of cases such as skydiving.In the case of skydiving, the same physiological changes (those that skydiving sup ose ly typica ly tri gers) occur in both someone who enjoys the experience and in someone who dislikes the experience (Schroeder, 2004).The same pattern of bodily changes can give rise to positive and negative valence.Valence is not then grounded in bodily perception.The independence claim holds.
I am not convinced by this line of reasoning.Assuming for the sake of ar ument that skydiving does tri ger the same pattern of bodily changes both in someone who enjoys the experience ("positive skydiving") and in someone who dislikes the experience ("negative skydiving"), there sti l is an alternative take on this kind of case.
Positive and negative skydiving can be seen as involving different bodily experiences, as the sa e pattern of bodily changes can give rise to different bodily percepts ("intero-ceptive percepts").Attentional mechanisms can straightforwar ly account for this phenomenon.Attentional modulation plays a key role during percept formation.Very roughly, in some influential accounts (e.g., Clark, 2013;Hohwy, 2013), during percept formation, attentional modulation (roughly, infer ing "precisions") determines which a ects of the incoming sensory signal are given more weight (i.e.amplified) and which a ects of the incoming sensory signal are ignored (see Clark, 2013;Hohwy, 2013).Several contextual factors influence such differential regime of weights assignment.For example, descending sensory expectations based on multilevel stored knowledge about the faced situation (see Clark, 2013;Hohwy, 2013).To take a typical example, from the same noisy stimuli, such as white noise, given different contextual cues, one can form either the auditory percept of a familiar song, or the percept of a conversation, or just random tones.In this sort of case, expectations tri gered by contextual cues determine which a ects of the incoming sensory ar ay are (subpersona ly) taken to be relevant and which a ects of the incoming sensory array can be taken to be just noise, relative to the expectation in question.There is no reason to deny that this process of expectation-based weight assignment also occurs in the case of percepts of the inner condition of the body.Now, let's assume that positive and negative skydiving involve different context-sensitive expectations about the "likability" of the coming experience.Let's say that, regarding the pro ects of skydiving, one person feels excited and that the other person feels afraid, so that the former expects the experience to be good, while the latter expects the experience to be bad.Considering that such differing expectations imply different assignments of precisions relative to the (purporte ly) same input stimuli, the percept that results from this process wi l be composed of different features, thus changing the configuration of the percept that wi l eventua ly be experienced.That is, given different kinds of context-sensitive sensory expectations, by differentia ly weighting different a ects of the sa e pattern of bodily changes via attentional mechanisms, the kind of interoceptive percept of the body formed in the case of positive skydiving can differ from the percept of the body formed in the case of negative skydiving.Thus, if this view is on track, in the case in question the pattern of bodily changes chara eristic of skydiving is not perceived as the same type of bodily state.Car uthers' ar ument fails then to e a lish its conclusion, simply because, in this account, skydiving does not count as a case in which the sa e bodily percept involves different valence values.Cases such as skydiving do not rule out then the possibility that valence, together with arousal, is also realized by representations in the interoceptive system3 .

Conclusion
As I attempted to show in this paper, Car uthers's version of NSS is pro lematic on its own, independently of the plausibility of competing theories of valence.The ar uments provided by Car uthers are not successful in showing that the independence claim holds.That is, his ar uments to the effect that valence is non-sensory fail to rule out the hypothesis that, together with arousal, valence might also be grounded in bodily, sensory representations.Car uthers's version of NSS needs to be then more thoroughly substantiated.Thus, even though NSS might be the case, Car uthers's ar uments are not compe ling in showing why this view might be on track.