Cognitive penetration and taste predicates : making an exception to the rule 1

1 This article has been elaborated in the framework of the project A Computational Dynamic Analysis of Public Debates on Politics, Aesthetics and Taste, funded by CONICYT/FONDECYT/POSTDOCTORADO/N° Proyecto 3180096. 2 Postdoctoral researcher at the Universidad de Valparaí so Instituto de Filosofía, Facultad de Humanidad y Educación, Serrano, 546, 2340000, Valparaíso, Chile. Email: davidbordonaba@gmail.com. ABSTRACT


Introduction
Food is becoming increasingly important in today' s societies. It has a direct influence on individuals, but it is also crucial in defining and shaping the different cultural identities that exist throughout the world. In this context, knowing what we eat is critical. For example, stamps warning about products with a high content of fats or sugar are becoming more frequent. In the same way, the origin and the traceability of the products we consume is becoming more and more important for people of different countries (Loureiro and Umberger, 2007;du Plessis and du Rand, 2012;van Rijswijk and Frewer, 2012;Sun, Wang and Zhang, 2017). It is also essential to know what can be eaten in a given culture, or even how something can or should be eaten, to resp ect sp ecific cultural and social policies. However, on certain occasions, knowing what we are eating can have consequences that we would not have previously susp ected. Sometimes, knowing what we are eating can lead us to disagree with other people, frequently with people belonging to a different culture, on the circumstances of application of the predicates we use to describe or evaluate the things we eat.
It is commonplace in the literature (Vahid, 2014;Stokes, 2015) to say that the thesis of cognitive penetration would have significant consequences for three areas within philosophy. First, on the philosophy of science, concerning theory-ladenness of perception and the consequences this may have for scientific theories. Second, on the philosophy of mind, in particular with the architecture of mind. Third, on epistemology, in particular about the epistemic role of perception.
The aim of this paper is twofold. First, to show that cognitive penetration is also relevant to the philosophy of language. I will argue that cognitive states such as knowing that the dish is made with whale meat or believing that the dish is made of cat meat can affect taste perceptions. In some instances, this influence can make sp eakers say things that make them disagree and, subsequently, negotiate on the circumstances of application of the "predicates of personal taste" (Lasersohn, 2005) they have used in those situations. 3 Specifically, I will contend that there are cases of cognitive penetration where ethical, social, or cultural rules affect our taste perceptions, and where sp eakers accommodate these conflicting contents as exceptions to the rule determining the circumstances of application of "tasty" . Following MacFarlane (2014), I will consider that "the TP Rule" (MacFarlane, 2014, p. 4) is the rule determining the use of taste predicates, 4 but I will defend that there can be cases in which sp eakers can consider exceptions to this rule.
Second, I will provide a theoretical framework to explain the proper dynamics of these cases. To do this, I will rely on two ideas. First, I will follow those theories that conceive context as common ground. According to Stalnaker (2002Stalnaker ( , 2014, the notion of common ground has been understood in two different ways: as the central component of the score of a language game (Lewis, 1979), and as a complex state definable in terms of the propositional attitudes of the participants in a conversation (Grice, 1989). In this paper, I will take the Lewisian route, arguing that the exceptions the sp eakers introduce to the TP Rule can be understood as a component of the score of the conversation. Second, I will apply Richard' s (2008) taxonomy concerning taste disagreements to accurately depict the attitudes and stances of the sp eakers involved in this type of situation.
The plan for the paper is as follows. In Section 1, I will briefly explain what is cognitive penetration, and I will defend that the central cases presented in this paper are instances of cognitive penetration. In Section 2, I will define the essential elements for a proper charact erization of the dynamics of the cases: the sp ecifications of the kinematics of the circumstances of application of taste predicates, and the rule of accommodation for exceptions. In Section 3, I will provide an example in which sp eakers accommodate conflicting contents as exceptions to the TP rule, and I will show how they add these exceptions to the common ground of the conversation. In Section 4, I will present Richard' s (2008) taxonomy concerning the attitudes sp eakers can have in a taste disagreement. I will defend that this taxonomy is a perfect fit to account for the cases that are the focus of this work. Finally, I will draw some conclusions.
14 Reykjavik, Iceland -Ready to try grilled minke whale skewers, the "Moby Dick on a Stick", as the dish is advertised by The Seabaron restaurant, a British tourist observes, "looks like a Turkish kebab." Taking a bite, he adds, "Tastes indeed like red meat and poor ethics." 5 Eating whale meat is certainly not common, at least not in many cultures around the globe. Whale meat may taste like red meat, as the tourist of the article says. However, for many people, this may not be, by far, the most relevant asp ect to describe or assess the grilled minke whale skewers served in the Seabaron rest aurant. As the tourist of the article also notes, whale meat not only tastes like red meat. It also tastes like "poor ethics. " Perhaps this is a somewhat enigmatic statement, but it also stresses something meaningful for this work. Sometimes, when describing or assessing how something tastes, we may not only have to take into account the physical elements responsible for the flavor of the dish.
Cognitive penetration is usually defined saying that cognitive states such as beliefs, thoughts, desires, expectations, concepts, or intentions can directly influence our perceptual experience. Although defining cognitive penetration in a theoretically satisfactory way can be a difficult task (see Machery, 2015;Stokes, 2014;Silins, 2016), it is helpful to adopt a definition that allows us to explore the different issues related with cognitive penetration. In this work, I will follow Silins's definition, according to which: (…) experience is cognitively penetrable just in case the following scenario is posible: two people are the same with respect to their sensory inputs, the state of their sensory organ, and the orientation of their attention, and they are still different with respect to what their experience is like, because of their beliefs, desires, or other cognitive states (Silins, 2016, p. 27).
In short, if two people eat the same, have similar sensory systems, and similarly direct their attention and, even so, they have different perceptions due to their beliefs or desires, then one can sp eak of cognitive penetration. Although the existing literature on cognitive penetration in taste perception is not too extensive, scenarios like the one depicted by Silins's definition are entirely possible.
Suppose that Eric and Erik are clones, and both have been living in the same environment for their entire lives. Their memories are also the same, with the only difference that Eric remembers having had a horse during his childhood. Suppose now that they are going to eat at a Japanese rest aurant and Erik orders basashi, a sashimi-like dish made from the meat of horse. The waiter brings the dish to the table and, both knowing what the dish is made of, they try it. In this situation, it is reasonable to think that the conditions set out in the definition are fitted. First, their sensory inputs are equal since the elements of the dish are the same for both. Second, we can suppose that their sensory systems are very similar in crucial resp ects. After all, both are clones, and both have lived all along in similar environments. Third, we can also suppose that the direction of their attention is very similar. In fact, it is easily imaginable that their attention is directed towards the same thing, since the variety of elements -the meat and the sauce-does not certainly distinguish sashimi-like dishes. However, we can imagine that their taste perceptions are different. Eric remembers having had a horse during his childhood, something Erik does not remember, and this can affect him by making his perception of the basashi unpleasant.
It might be thought that for most people, these types of issues do not have much relevance in determining taste perceptions. Perhaps, for most people, the only things that count most of the time to say that something is "tasty" are the ingredients of the dish and the exact amount of each of them. However, according to Burnston's (2017) "scaling argument, " "the more high-level, categorical, socially-mediated, and learning-dependent a percept is, the more likely it is to be the result of CP [cognitive penetration]" (Burnston, 2017, p.1).
Taste perception is unquestionably socially-mediated. What we love or hate to eat depends, among other things, on the society or culture to which we belong. Although there may be certain overlaps due to geographical proximity, for example, in Mediterranean cultures, the use of olive oil is widespread, the most typical dishes of many cultures are very different. Consider, for example, the differences between Spanish and Syrian cuisine. Besides, taste perception is also learning-dependent. Although first-hand knowledge is sufficient for many people to consider someone' s tastes as reasonable, many people try to refine their taste through learning (see Smith, 2007;MacFarlane, 2014). In many situations, the more knowledge I have of the object I taste, the more pleasure I can get of it. Although someone might think this is elitist, the truth is that taste education is reaching more and more places and people every day. Consider, as proof of this, the enormous number of tasting courses on offer today. The wine tasting courses are undoubtedly the most popular since long ago, but other options comprise beer, coffee, chocolate, oil, cheese, or whiskey tasting courses.

15
So, cognitive penetration where ethical, social, or cultural rules affect our taste perceptions is something we have to look at carefully. The influence that some of these subjects can have on taste perception has already been observed in the literature: However, the modern problem is different. Cultural influences on our behavioral preferences for food and drink are now intertwined with biological expediency that shaped the early version of the underlying preference mechanisms. In many cases, cultural influences dominate what we eat and drink. Behavioral evidence suggests that cultural messages can insinuate themselves into the decision-making processes that yield preferences for one consumable or another (McClure, 2004, p. 379).
How culture and society influence our taste preferences has been studied, for example, in product labeling (Wansink et al., 2000), or product brand (McClure, 2004). Although this is a topic of great interest, in this paper, I will focus on a subject little explored in the literature about cognitive penetration in taste perception. I will examine those cases in which the influence of ethical, social, or cultural rules affects taste perceptions, making that sp eakers utter conflicting contents that lead them to disagree and eventually to negotiate on the circumstances of application of the predicates they have used. Therefore, the ultimate interest of the article is not in the taste perceptions themselves, but in the predicates the sp eakers use to describe or express those perceptions.

Scorekeeping in a tasting game
After showing that there can be cases in which ethical, social, or cultural rules cognitively penetrate our perception of taste, the objective now is to propose an explanation of the proper dynamics of these types of cases, placing particular emphasis on the circumstances of application of taste predicates.
MacFarlane (2014) contends that the rule determining the use of a predicate such as "tasty" is the Taste Pleasing (TP) Rule: "If you know first-hand how something tastes, call it 'tasty' just in case its flavour is pleasing to you, and 'not tasty' just in case its flavour is not pleasing to you" (MacFarlane, 2014, p. 4). I agree with MacFarlane that the rule that determines the use of taste predicates is the TP Rule. However, as already shown, there may be cases where, because of cognitive penetration, there are other rules that may be relevant. As said, I will follow MacFarlane in considering the TP Rule as a use rule, but I will accept the possibility that sp eakers consider exceptions to TP. I will treat the set of exceptions that are relevant in the context as a component of the score of the conversation. Lewis (1979) stated that just as the score in a baseball game allows you to know the state of the game, the score of a conversation would allow you to know the state of a given conversation at each stage. A baseball game is an attainable scenario because we can depict it, as Lewis does, as a septuple of elements: (2 visitor /0 home /top/7 inning /2 strikes /3 balls /2 out ) I would add two other elements to the septuple suggested by Lewis: i) what bases are loaded and the position in the batting order occupied by the batter. It is not the same to have a runner on first base and the pitcher batting than to have the bases loaded when the cleanup hitter is in the box. So, the score of a baseball game would be the following nonuple: (2 visitor /0 home /top/7 inning /1&3 occupiedbases /5 hole-at-bat /2 strikes /3 balls /2 out ) Maybe Lewis did not explicitly add these elements to the score because they are recognizable by merely looking directly at the game, at least what bases are loaded. In the same sense, there are features of a conversation (e.g., who is the sp eaker) that are recognizable just joining or being part of a conversation. However, others require certain knowledge of the conversational score, for example, what presuppositions are part of the common ground at a given moment or, in our case, what exceptions to the TP Rule sp eakers are considering.
Be that as it may, the general idea is that a quick look to the score is enough to know the state of the game at a particular stage. However, knowing the different elements making up the score of a conversation at a given point could be more challenging than in the case of a baseball game. Nevertheless, as Lewis stresses, there are certain points of similarity between these two scenarios, the main ones being: • Both scores can be represented by a set of components that are abstract entities. In the case of the score of a baseball game: the number of runs, the inning, or the number of outs, balls, and strikes at a given moment.
In the case of a conversational score: sets of presuppositions, permissible or impermissible courses of act ion, rankings of salience for definite descriptions, standards of precision, or the exceptions to the TP Rule that are relevant. • Both the score of a baseball game and the score of a conversation "evolves in a more-or-less rule-governed way. " (Lewis, 1979, p. 345). When a given sp eaker says that something is "tasty, " the TP Rule is the rule that entitles her to use the predicate. However, in certain situations, other rules can make that the sp eakers consider exceptions to the TP Rule. Here is the definition of the kinematics of the circumstances of application of taste predicates: Specifications of the kinematics of the circumstances of application of tast e predicates: Initially, the rule that entitles 16 sp eakers to say that something is tasty is the TP Rule: call something "tasty" if its flavor is pleasing to you and you know first-hand how it tastes. After that, if between t and t' sp eakers behave in manner m (i.e. uttering a conflicting content), one or more exceptions to TP can be considered to be relevant in the context. Speakers can consider certain exceptions to the TP Rule. The participants in the conversation will accept these exceptions as part of the common ground if the assertion of the sp eaker is to be acceptable. Let us consider a context in which TP is at stake. Speaker A and Speaker B are eating at A's house. Speaker A offers Speaker B some food, and both of them say that the dish is tasty. At some point in the conversation, Speaker A reaffirms his position by an utterance that adds new information, updating in this way the context. In other words, through a "manifest event, " i.e., through "(…) something that happens in the environment of the relevant parties that is obviously evident to all" (Stalnaker, 2014, p. 47), in this case, an utterance that adds new information to the context, he tries to make the asserted content be part of the common ground, while TP is the rule at stake. At this point, suppose that Speaker B expresses disagreement with Speaker A because Speaker B considers the new content and TP as part of the common ground to be jointly incompatible. Now, Speaker A can accept as common ground both the new content and TP if she accepts the information added by the new content as an exception to TP. In other words, sp eakers can "accommodate" (Lewis, 1979) these new conflicting contents as exceptions to the TP Rule as the conversation evolves. Here is the definition of the rule of accommodation for exceptions: Rule of accommodation for exceptions: If at time t one sp eaker says something (a conflicting content) that produce a disagreement with other sp eaker, sp eakers can accept these new content as an exception to the TP Rule. If the exception was not present before t, then the exception can be taken as part of the common ground at t.
In the above example, one option is that Speaker A concedes to Speaker B that the exception is part of the common ground. 6 If that were the case, a new component would arise in the score of the conversation, the set of exceptions that are relevant for saying that something is "tasty. " From that moment, knowing the circumstances of application of "tasty" in that context will imply knowing what exceptions the participants in the conversation have accepted as part of the common ground.
It may be argued that accommodation is not a general way in which sp eakers account for those changes that arise in the course of a conversation. However, on the face of the number of changes and the different kinds of change a context might undergo, I will take accommodation as a necessary feature of any disagreement involving taste predicates.

Tastes, rules, and exceptions
Considering the set of exceptions as a component of the score of the conversation allows us to account for those cases in which ethical, social, or cultural rules make an impact on the circumstances of application of taste predicates. To have first-hand knowledge of the object of which sp eakers are expressing their gustatory preference is key in many situations, but, as the conversation goes, sp eakers can say something that makes sp eakers disagree, being possible to accommodate these conflicting new contents as exceptions to TP.
In the remainder of this section, I will present an example to illustrate the ideas defended so far in the paper. However, before continuing, a proviso is necessary. The examples depicting taste disagreements (Glanzberg, 2007;Huvenes, 2012;Lasersohn, 2005;López de Sa, 2008MacFarlane, 2014) typically consist of one sp eaker saying p and the other sp eaker saying q (where q implies not-p). Such examples make it impossible to illustrate those cases in which the sp eakers consider exceptions to TP. The example-style I will use can be found in other authors (Stevenson, 1944;Schaffer, 2011;Blome-Tillmann, 2014). Like them, I will present larger and more complex pieces of discourse to show the sp ecific kinematics of these cases. Now, consider two sp eakers, Mike and Pedro, eating the typical dish of the Cañete gastronomic festival: 1) Pedro: Umm, it's really tasty. It's been a long time since I've tried it. 2) Mike: Yes, it's really tasty. What kind of meat is in the dish? 3) Pedro: You don't know? It's cat meat. 4) Mike: Cat meat? Are you kidding me? This is the most disgusting thing I've ever tried.
Pedro and Mike agree on (1) and (2) that the dish is tasty: the flavor of the dish is appealing for both sp eakers, so both of them are willing to express their approval of the dish by calling it "tasty. " Suppose that Mike presupposes that it is rabbit or chicken meat, and Pedro presupposes that it is common ground that the dish is made of cat meat. Pedro could think that someone who goes to the Cañete gastronomic festival knows that one of its most prominent features is that the dishes are made with cat meat. Mike' s question in (2) indicates that he does not know what meat is in the dish, and then it is common ground that he does not know the type of meat in the dish. After that, since Pedro knows that Mike does not know which type of meat is, Pedro asserts (3). Now it is common ground that it is cat meat. Once Pedro updates the context in (3) adding the information "cat meat, " Mike has two options.
On the one hand, he can accept as common ground both that the dish is made of cat and that TP determines the circumstances of application of "tasty. " On the other hand, he can refuse to accept either as common ground. In this case, Mike opts for this second option, expressing his disapproval of the dish in (4). If Mike had not said anything, Pedro would have taken for granted that he accepts as common ground both that the dish is made of cat and that TP is the only element determining the circumstances of application of "tasty. " According to Stalnaker (2014), sp eakers can realize that the conversational context is defective, for example, when "(…) it becomes manifest that you are presupposing something that I don't, or didn't, believe" (Stalnaker, 2014, p. 47). He identifies three ways 7 for sp eakers to accommodate these situations. Following with the example, after Mike's utterance of (4), here we have a situation when one sp eaker, Pedro, has several options to accommodate this "defect" of the conversation. First, he may accommodate it by changing his belief and taking it to be common ground that eating cat meat in the Cañete gastronomic festival is disgusting. In doing so, Peter would accommodate the new conflicting content "cat meat" as an exception to the TP rule. One consequence of this move is that the exception to TP would be relevant in the context, and the circumstances of application of "tasty" would no longer depend exclusively on the TP Rule, but instead on TP plus the exception. We could summarize the new circumstances of application as follows: call something "tasty" if its flavor is pleasing to you, and you know first-hand how it tastes unless you are eating the meat of a pet. 8 However, Pedro can also try to challenge Mike' s point. Consider the following possible development of the conversation: 5) Pedro: Wow, this is unbelievable. You knew you were going to come to the festival but you didn't even check what kind of festival it is? Ok, I'll tell you the story. The festival is a hallmark for African-Peruvians. It's devoted to Santa Efigenia, an Ethiopian black virgin, one of the people responsible for the spread of Catholicism in Africa, and one of the few black representations of the Virgin in the world. I don't know exactly why cats are eaten in the festival, but I don't think it's so important. The festival is now a distinguishing mark for Peruvians with African ancestry, not a majority in Peru, as you might guess. 6) Mike: That story makes all the difference. I knew nothing about the importance and significance of the festival. Now, I understand why eating cat meat doesn't bother you. In fact, I'm beginning to think that it's not entirely appropriate to bluntly say that the dish is disgusting. In the end, eating cats is just a way to reinforce group cohesion in a discriminated minority. I think I'll have to give it another chance.
In (5)-(6), another exception has been introduced, in particular an exception to the first exception. Now, the circumstances of application of "tasty" could be rendered as follows: call something "tasty" if its flavor is pleasing to you and you know first-hand how it tastes, unless it is pet meat, with the exception that it forms a significant part of a tradition that favors social cohesion. This last change has shifted again the conditions under which it is permitted to say that the dish is "tasty. " The exceptions that sp eakers introduce as the conversation evolves can be of two kinds: primary and secondary exceptions. A primary exception is an exception to TP. In the example, it was "unless it is pet meat that you are eating. " Secondary exceptions are exceptions to primary exceptions. In the example, "(…) with the exception that it was a significant part of a tradition that favors social cohesion. " Note that if the primary exception had been a different one, for example, "unless it is chimp meat that you are eating" , there would probably have been fewer possible secondary exceptions. Had the primary exception been "unless it is human meat that you are eating" , secondary exceptions would be admissible in only very few contexts. So, it can be said that primary exceptions coerce the range of possible secondary exceptions in a given situation.
To sum up, this example shows how the sp eaker's stance change as the conversation evolves. In this process, sp eakers add new information to the context, and these new pieces of information are accommodated as exceptions that are relevant for saying that the dish is tasty. Speakers agree at the beginning, but after Pedro adds the conflicting information "cat meat, " Mike disagrees, trying that Pedro accepts an exception to TP. After that, Pedro adds the accordant information "social cohesion, " which adds a new exception. Mike finds this reasonable, and in the end, both of them accept TP, and both exceptions as common ground.
It should be remembered that the disagreement is not about the exceptions that need to be considered to say that the dish is tasty. The set of exceptions is a component of the score of the conversation, and as such, it can evolve depending on the conversational moves of the sp eakers. The disagreement between Pedro and Mike is a disagreement on whether the typical dish of Cañete's gastronomic festival is tasty or not, as illustrated by the fact that, in the end, both sp eakers agree that the dish is tasty. However, we can ask ourselves, is it correct to say that Peter and Mike end up agreeing at the end of the conversation? Perhaps "agreement" is not the term that 18 best describes the situation reached by both at the end of the conversation. Maybe, we should not say that they agree, since although it is true that for Peter the dish is tasty, it is far from clear that for Mike it is.
4. Agreement, acceptance, deficiency and intolerability: Overcoming the agreement/ disagreement dichotomy The example in the previous section is esp ecially illuminating because it exposes a situation that is difficult to account, at least taking into consideration most of the proposals available on the literature on taste disagreement. A large part of the works has framed disagreements on taste as "faultless disagreements" (Kölbel, 2003;Schaffer, 2011;MacFarlane, 2014). A faultless disagreement is usually defined by saying that two sp eakers are in a faultless disagreement when: 1) Speaker A affirms or believes p; sp eaker B affirms or believes q (where q implies not-p) 2) None of the sp eakers has made a mistake in stating or believing p or q.
For years there has been much ado about how a disagreement where neither of the sp eakers is in error is possible. If it is a disagreement, then one of the parties will have made a mistake. If no one has made a mistake, then it is not a disagreement. These two contradictory but equally plausible ideas have made faultless disagreement a theoretical paradox to some extent. For this reason, some authors (Glanzberg, 2007;Stojanovic, 2007;Buekens, 2011;Rovane, 2012) have rejected the possibility of such disagreements, while other authors (Lasersohn, 2005;López de Sa, 2008;Kölbel, 2003Kölbel, , 2009Schaffer, 2011;Egan, 2014;Huvenes, 2012;Kompa, 2016;Wyatt, 2018) have tried to accommodate the phenomenon by developing theories designed explicitly for such a task.
The question of whether we should dismiss the idea of faultless disagreement or whether, on the contrary, we should accept it and try to explain it will not be directly relevant to the dialectic I pursue in this work. Instead, what matters here is to emphasize that most of the literature on taste disagreements only uses the notions of agreement and disagreement to represent the attitudes the sp eakers have in conversations on matters of taste.
Nevertheless, this is a problem because these two notions are not sufficient to represent some of the attitudes that sp eakers may have in this type of conversation, esp ecially if we consider examples such as the one represented in (1)-(6). As we have seen, when sp eakers update the context by adding new information that has an impact on the circumstances of application of taste predicates, one option for accommodating this new information may be to accept it as a primary or secondary exception. However, the attitudes sp eakers may adopt during or after that happen can hardly be categorized under the agreement/disagreement dichotomy. To account for this deficiency, I will apply Richard' s (2008) taxonomy to the dynamic framework presented in the previous sections. This will allow us to give a proper description of the attitudes sp eakers adopt in the type of cases presented. According to Richard: There are two ways in which I could find your valuing x as cool or otherwise to be appropriate. First of all, our perspectives could simply agree on x-we both, say, think x is cool. Or it could be that while we value x differently I acknowledge your perspective as one way to think, a way of valuing I have no reason to oppose that I think you ought to accept. I acknowledge that your attitude towards x is perfectly appropriate-I just don't share it. I will call these attitudes towards your valuing x as cool (or not cool, or handsome, or not sexy, or . . . ) agreement and acceptance respectively (Richard, 2008, p. 130).
Likewise, Richard (2008) unfolds in two the idea that we can disagree with someone: There are also two ways in which I could find your valuing x in a certain way inappropriate. First of all, even if I acknowledge your perspective and thus don't find your evaluation in error, I may differ with you over whether there is a reason to find x is cool you ought accept. If I do, I find your valuing x liable to reproach; I, as I shall put it, find valuing x as you do deficient. In this case I may while acknowledging that yours is one acceptable way to go through life, try to convince you that there is a better way. Secondly, I may not just find your valuation of x deficient; I may think your perspective intolerable, one which does not constitute an aceptable way of looking at things. I may, that is, be intolerant of your judgment, insofar as it is a manifestation of a perspective on the matter at hand which I find unacceptable (Richard, 2008, p. 130).
In short, Richard (2008) makes more complex the panorama of available options to describe the attitudes or stances that the different parts can take in a taste disagreement. Most of the literature dealing with this type of disagreements uses only the ideas of agreement or disagreement to describe the positions that can be taken. However, when things are not so simple, for example, when sp eakers consider exceptions to the TP Rule, we need more options because the attitudes of the sp eakers cannot be described simply by saying that they agree or disagree.
Consider our toy example again. As already mentioned,

19
at first, both sp eakers seem to agree on thinking that the typical dish of the Cañete gastronomic festival is tasty. However, after Pedro updates the context in (3) with the conflicting information "cat meat, " Mike expresses his disagreement in (4) since he cannot accept both that the dish is made of cat and that TP is the only element determining the circumstances of application of "tasty. " Up to here, the notions of agreement and disagreement seem to be sufficient to describe the attitudes of both sp eakers. However, after Mike says he finds it intolerable to call a dish made with cat meat tasty, Pedro explains to Mike the cultural and social background behind the celebration. Once Pedro explains this, and Mike understands the importance and meaning of the celebration, his attitude changes. Now, eating cat meat no longer seems intolerable, but acceptable. Understanding that the celebration is a way to reinforce group cohesion in a discriminated minority, Mike has no reason to oppose Pedro. Although he values the dish differently, after all, he does not think it is tasty, or at least not to the same extent as Pedro does, Mike recognizes Pedro's position as one he has no reason to oppose. He understands it, but he does not share it.
In sum, Richards's taxonomy gives us the conceptual tools to describe in an appropriate way the attitudes that sp eakers have in those situations in which they consider exceptions to TP. The resulting theoretical framework is rich enough to illustrate in all its complexity the dynamics of examples such as (1)-(6) (see Table 1).

Conclusion
In this paper, I have shown that cognitive penetration has relevance not only to the three traditional areas in which the phenomenon has been studied -philosophy of science, philosophy of mind, and epistemology -but also to the philosophy of language. I have argued that cognitive states related to ethical, social, or cultural rules can affect taste perceptions, and, depending on the conversational moves of the sp eakers, this can lead to a change in the circumstances of application of the taste predicates sp eakers use to describe or express their taste perceptions. Specifically, sp eakers can consider exceptions to MacFarlane's (2014) TP Rule. I acknowledge that TP is the rule determining the circumstances of application of taste predicates, but I have shown that in some instances, sp eakers can consider exceptions to the TP Rule.
Then, to explain the proper dynamics of these cases, I have developed a theoretical framework build upon two elements. First, I have used the idea of context as common ground, more sp ecifically the Lewisian version of the idea. If one sp eaker updates the context with new information, while still upholding TP, the other party can express disagreement if he or she is unwilling to accept the new content and TP as common ground. One way to accommodate this is to add the content to the common ground as an exception to TP. I have provided the sp ecifications of the kinematics of the circumstances of application of taste predicates, and the rule of accommodation for exceptions to provide a theoretical framework to explain the proper dynamics of these cases. Second, I have used Richard's (2008) taxonomy to properly illustrate the different attitudes sp eakers can have in these cases.

Stage of the conversation
Information in the common ground Circumstances of application Pedro's attitude Mike's attitude (1)-(2) The TP Rule TP Agreement Agreement Conflicting information: "cat meat" (3)-(4) The TP Rule Cat meat TP Agreement Disagreement Accordant information: "favors social cohesion" (5)- (6) The TP Rule

Cat meat
Favors social cohesion TP + E1 + E2 Acceptance Acceptance