Testimonial Kinds: The Source Factor

: In this article, I argue that the nature of testimony necessitates that we distinguish between testimonies that are based on the informant's sense perception, inference, or on a longer testimonial chain. I further argue that this distinction has epistemic signi ﬁ cance, in that it helps us better understand how reliable certain classes of testimonies are and how reliable certain individuals are, based upon the epistemic source that their testimony is ultimately grounded in. I begin the article, in Section 1, by drawing attention to the existence of several di ﬀ erent terms for testimony found within the Islamic philosophical tradition. In Section 2, I argue that the essential di ﬀ erence between these terms has to do with the epistemic source that grounds the speaker ’ s testimony. In Section 3, I move on to explicate how this distinction helps us better understand the reliability of testimony and how it impacts our evaluation of the reliability of di ﬀ erent speakers in di ﬀ erent contexts. In Section 4, I use this distinction to challenge the necessary condition of speaker competency. Finally, in Section 5, I use the distinction to undermine the belief-transmission view in the epistemology of testimony.


Introduction
The thirteenth-century Muslim jurist, Shihāb al-Dīn al-Qarāfī (d. 684/1285), famously began his multivolume work, al-Furūq, explicating the essential difference between several kinds of statements (akhbār). 1 Al-Qarāfī's work, whose title translates literally as The Book of Differences, focuses on the subtle distinctions between foundational principles in the Islamic tradition. He opens the first chapter by informing his readers that he spent eight years in search of the essential difference between two specific types of statements (akhbār), shahāda and riwāya, and was only met with replies highlighting the various norms associated with each category. 2 He rightfully points out that these norms ultimately stem from some essential difference and therefore did not really answer the question.
Although al-Qarāfī and others in Islamic intellectual history have advocated for a solution to this puzzle, I believe that the essential difference has yet to be plainly articulated. 3 In the rest of this article, I will argue that the essential difference between shahāda and riwāya, as well as other species of statements (akhbār), is the source of knowledge upon which the speaker rests. This is an argument about the nature of testimony, that it comes in many forms, and that these technical terms from the Islamic tradition are tracking these differences. This argument is not merely historical, for these categories exist irrespective of what any given philosophical tradition might call them.
The second argument concerns the epistemological value of this distinction. As I will argue in the following sections, given the unique external nature of testimony, the epistemic source of the content being communicated is key to its analysis. In other words, this distinction in species is not just an interesting relic of intellectual history; rather, it is an intuitive distinction that helps us make sense of how the reliability of testimony is thought about and managed in our intellectual lives.
Section 2 is dedicated to explicating the division itself. In Section 3, I will be showing how this division impacts the reliability of the speaker and thus impacts justified belief and knowledge. In Section 4, I will show how this division helps us better understand the hearer and speaker competency condition. Finally, Section 5 will show how this division helps undermine the belief-transmission view in the testimony literature.

The Epistemic Source of the Speaker
The three most common sources of knowledge mentioned in the Islamic intellectual tradition are the intellect (al-ʿaql), the senses (al-ḥawāss), and true statements (al-khabar al-ṣādiq). 4 While some Muslim jurists, theologians, and philosophers mentioned other sources, such as introspection (al-wijdān) or Divine inspiration (alilhām), the first three are always included in the taxonomy. This is most likely due to their centrality in topics related to theology and jurisprudence. 5 In the anglophone tradition, the term testimony is used to refer to a source of knowledge akin to sense perception, reason, memory, and introspection. 6 In contrast, in the Islamic tradition, statement (khabar) is used to refer to that same source. 7 The category statement (khabar) has another perhaps more fundamental role in the Islamic philosophy of language literature, as it is often juxtaposed against performatives (inshāʾ). However, as a matter of usage, the term khabar is used both to signify the linguistic category and the epistemic source itself.
Thus, the sources of knowledge are articulated based on the vehicle through which knowledge reaches the subject. In the case of reason, it is through the intellect (al-ʿaql) of the subject. In the case of sense perception, it is through his senses (al-ḥawāss). Finally, in the case of testimony, it is through the true statements of others (al-khabar al-ṣādiq). Therefore, the traditional Islamic taxonomy of sources of knowledge, the intellect (al-ʿaql), the senses (al-ḥawāss), and true statements (al-khabar al-ṣādiq), end up being analogous to reason, sense perception, and testimony.
According to traditional Islamic epistemology and philosophy of language, there are various species or kinds of statements (akhbār). Given the variety of testimonial kinds, Islamic scholars often sought to make sense of the differences between them. It is against this backdrop that shahāda and riwāya are often contrasted against one another and analyzed.
The context Muslim theologians, jurists, and philosophers had in mind when discussing shahāda is formal court cases or formal cases being brought before some type of governmental ruler or bureaucrat. Similarly, when they discussed riwāya, they had in mind the formal practice of transmitting Prophetic statements (ḥadīth). This seems to have resulted in their focusing on formal differences, such as the scope of the speech act, as opposed to highlighting the epistemic origins of each species.
However, if we isolate the formal contexts from each term, the differences referred to earlier do not hold. While it is true that shahāda is formally associated with a specific speech act and attitude on the part of the speaker and entails a certain scope, it may lose these features in more informal and natural settings. Similarly, formal instances of riwāya are associated with certain norms and have certain implications when it comes to the context of Prophetic testimonies, but most, if not all, said restrictions might fade away in more informal and natural instances. Given this difference between formal and natural instances of shahāda and riwāya, it cannot be the case that the aforementioned features are responsible for the initial division, nor are they sufficient in explaining all the differences in norms associated with each term.
Instead, I propose that the essential difference between shahāda and riwāya is found in how each species is epistemically grounded in the speaker. If the epistemic source of the speaker is their own sense perception, it is an instance of shahāda, which we can translate as sensory testimony. In contrast, if the epistemic source of the speaker is itself testimonial, then it is an instance of riwāya, which we can translate as transmission. If the epistemic source of the speaker is inferential, neither of these terms is used.
In the Islamic tradition, testimonies that are substantially grounded in the inference of the speaker are referred to by a few different names and are not usually compared and contrasted to sensory testimony (shahāda) and transmission (riwāya). Nevertheless, they are flagged as their own species, particularly in the formal case of a jurist or expert authority issuing a legal opinion. For this reason, one of the most common terms used in the Islamic tradition for this species is opinion (raʾy). Here, like in the previous two cases, the essential difference is the source of knowledge and the cognitive process of the speaker. The legal opinion being issued by a jurist or expert authority is generally going to be the result of some type of inferential process on their part.
Thus, in the Islamic tradition, sensory testimonies (shahāda) are a species of statements (akhbār) that are grounded in the speaker's sense perception. Transmissions (riwāya) are a species of statements (akhbār) where the speaker has received the content testimonially. Opinions (raʾy) are a species of statements (akhbār) that are substantially grounded in the speaker's inferential process.
Having now introduced the central thesis of this article, it is helpful to return to the initial taxonomy of sources to see how this distinction coheres with the broader system. Some Muslim theologians sought to explain the relationship between reason, sense perception, and testimony in a way that can help us better make sense of the distinctly unique role of testimony. They explain this relationship by employing a classic division in how each source relates back to the subject under epistemic evaluation. According to this division, sources of knowledge are either external or internal to the subject. As for sources that are internal to the subject, they are further divided according to whether they are associated with that which is doing the perceiving (the rational soul), or they are an instrument of that which is doing the perceiving (the senses). According to this division, testimony is distinct in its external nature. 8 Testimony is indeed fundamentally different than other sources of knowledge because it is the only source that significantly relies on another epistemic agent. Other sources of knowledge, such as sense perception or reason, rely on non-agential external stimuli or rely on other epistemic agents in a non-significant manner. Therefore, any epistemic analysis of testimony as a source of knowledge requires an analysis of not only the hearer but also of the speaker.
There is a significant overlap in these distinctions between the Islamic tradition and contemporary anglophone epistemology. Recent work on social epistemology, particularly on the social aspects of the epistemology of testimony, has highlighted the importance of investigating the epistemology of social interactions. 9 The commonly accepted narrative is that epistemologists have long focused on evaluating the doxastic attitudes of individuals in abstraction and have only recently taken a social turn. According to this narrative, contemporary anglophone epistemology looks beyond the epistemic subject in question to factors  8 Al-Taftāzānī, Sharḥ al-ʿAqāʾid, 113. 9 See, for example, Gelfert, A Critical Introduction to Testimony, 8; Goldman and Whitcomb, Social Epistemology. external to him. When these external factors include other subjects, the social elements of epistemology are brought front and center. Islamic epistemology seems to have anticipated many of these points and built upon them in interesting and unique ways.
Furthermore, there is an overlap between the testimonial-kinds-thesis being advocated for in this article and the anti-individualistic epistemology advocated for by Sanford Goldberg, which extends the area of investigation beyond the hearer to the reliability of the cognitive process of the subject's informant. 10 While it would be anachronistic to talk about traditional Islamic epistemology as being externalist, the various distinct species of testimony track the different cognitive processes of the informant. Given the nature of testimonial exchange, a hearer's knowledge, justification, and other epistemic properties are directly impacted by the epistemic properties of the speaker. The epistemic properties of the speaker are, for their part, a result of their own cognitive process and epistemic source. In this way, the central thesis of this article is thoroughly anti-individualistic. 11 When it comes to reason and sense perception, the subject's belief is ultimately grounded in either their own sense perception, reason, or some hybrid of the two. However, when it comes to purely testimonial knowledge, the beliefs that are a result of the propositional content being communicated from the speaker to the hearer are not ultimately grounded in the hearer's sense perception or reason. For it to be considered purely testimonial, it must significantly rely on the propositional content communicated from the speaker to the hearer. This much should be uncontroversial.
More controversial is the claim that this conclusion only gets half of the answer right. Given the social nature of testimony, there are at least two epistemic agents involved in the testimonial exchange. Thus, a thorough epistemic analysis of the testimonial exchange must also investigate the epistemic origins and cognitive process of the content being communicated by the speaker. This, I contend, is an important distinction that helps us better understand the complexity of testimony in our lives.
Before moving on to the next section, there is one possible objection that deserves engagement. In the case of transmission (riwāya), a speaker conveys content they themselves received testimonially. Therefore, by the very logic used to craft the initial division, transmissions (riwāya) themselves must ultimately be grounded at the end of the chain in a person's sense perception or reason. Thus, the transmission (riwāya) category loses any importance and explanatory significance.
My response to this objection is to concede the point that testimonial chains do ultimately, as a matter of fact, bottom out non-testimonially. 12 However, what is not conceded is that this fact entails the vacuousness of the category. Just because transmissions (riwāya) will eventually be grounded in some speaker's non-testimonial source does not mean that a hearer will always be able to access the first speaker in the testimonial chain.
Furthermore, even if transmissions (riwāya) are grounded upstream in other sources of knowledge, the epistemic analysis of the entire communicative act must include one subset of that act, the transmission (riwāya). 13 Even if one were to concede that a transmission (riwāya) is only as strong as the original source's testimony, the more proximate link in the testimonial chain could render the entire testimony weaker for any number of reasons. 14 This is precisely the case in which ḥadīth scholarship flourished. Islamic scholars were  10 Goldberg, Relying on Others. 11 The anti-individualistic component of the thesis coheres well with externalism. However, my objective here isn't to make a case for externalism; rather, it is to highlight the importance of the epistemic source of the speaker. How one chooses to make sense of that according to their theory of knowledge is beyond the scope of this article. 12 There is a case to be made for a type of testimony that does not bottom out in sense perception, reason, or any of the other wellknown sources of knowledge. However, that case relates to Divine testimony and will require a longer analysis and explication which is beyond the scope of this article. 13 This line of objection is undesirable for another reason. Its logic is transferable to the entire category of testimony itself. For one could argue, given that testimonial chains are ultimately grounded in other sources of knowledge like reason or sense perception, that testimony itself is a useless and vacuous category. This is of course not the intent of those who would argue against it since they would like to maintain the category of testimony while arguing that there is nothing epistemically useful in the rest of the classification. 14 I thank one of my anonymous reviewers for help in articulating this point. certain of the reliability of Prophet Muhammad; therefore, their epistemic method focused on evaluating transmitters further down the testimonial chain.
Up until this point, I've limited myself to light historical and conceptual analysis of testimonial kinds. The first two sections have served the purpose of laying out the contours of the problem, the area of investigation, and sketching out a way forward. I'll now be turning my attention to making an epistemic case for these distinctions, as well as showing how they can aid epistemologists in making sense of testimonially based exchanges. The remaining sections will focus on how these differences can help us analyze speaker reliability, competency, and respond to the belief-transmission view in the literature.

The Reliability of the Speaker's Epistemic Source
Building off the previous section, I propose that this testimonial-kinds-thesis has epistemic significance. I call this epistemic significance the Testimony Grounding Principle: TGPB: The reliability of H's testimonially-based justified belief in p depends in part on the epistemic source of the speaker.
TGPK: The reliability of H's testimonially-based knowledge in p depends in part on the epistemic source of the speaker. 15 It is perhaps useful to think of this principle as related to Jennifer Lackey's three classes of inductively based positive reasons. 16 In addition to contextual features, kinds of reports, and types of speakers, we could add the epistemic source of speakers as a class of reasons that helps us distinguish between reliable and unreliable testimony. Whether these reasons are about the class itself or about a given speaker and their relationship to that class, the epistemic source of the speaker gives us a better understanding of the reliability of the testimony under evaluation.
In this section, I use Keith DeRose's famous bank case, although slightly different than how he conceived it. 17 Instead of using the case to show how our intuitions about knowledge change based on high-stakes cases, I will use my version of the case to showcase how a speaker's reliability and, thus, their justification is impacted by testimonial kinds. BANK CASE D: Mike and Lucy are driving back to campus on Friday afternoon after a short vacation with their parents. Before getting on the highway, Mike wants to stop at the bank and deposit some cash. Unfortunately, given the time of year and the out-of-order ATM, the bank is very busy. He is just within his lease's grace period, so he needs to make the deposit by Saturday, or else he risks a heavy fine. Lucy, on the other hand, needs to get back to campus as soon as possible so that she doesn't miss an important student government meeting. After noting that the bank will be open on Saturday, Lucy encourages Mike to make the deposit then. Mike has no previous knowledge about his bank opening on Saturdays, but he is aware that this Saturday is a holiday, and he considers Lucy to be a reliable testifier.
DeRose's bank case was meant to highlight how context and practical concerns might shift the stakes for us as epistemic agents. BANK CASE D is no different. The stakes for Mike are high; he risks a heavy fine, perhaps even a bad credit rating. Given these stakes, it seems that Lucy's testimony, despite her general reliability, is not going to be sufficient. This is because Lucy's general reliability includes cases of testimony that are simply too varying and heterogeneous. How did Lucy come to know what she was testifying to? Is it based on her having heard it from the bank manager, or is it based on experience and induction?
This heterogeneity in testimonial kinds impacts Lucy's reliability. Lucy's reliability through transmission (riwāya) from the bank manager is not going to be the same as her reliability through sensory testimony (shahāda) or through opinion (raʾy). I don't mean to say that as a class of testimonies, one is always stronger  15 I've articulated the principle twice to account for the differences between justified belief and knowledge. 16 Lackey, Learning From Words, 181-3. 17 DeRose, "Contextualism and Knowledge Attributions." than the other; I only mean to say that, among other factors, the epistemic origins of the testimony impact its reliability.
Given this impact on Lucy's reliability, Mike's justification shifts accordingly. Whether or not Mike has justified belief or knowledge ends up being impacted by the cognitive process of the speaker, which is represented by these testimonial kinds. This is true irrespective of whether one adopts an internalist, externalist, or hybrid approach.

The Role of Competency in Reliability
There is almost universal agreement in contemporary epistemology that a speaker's sincerity and competency directly impact their reliability. 18 While tracking sincerity might be straightforward, competency is more complicated. Is the competency of a speaker only related to the subject area of the content being transmitted? How exactly do non-specialists track the competency of a specialist? Are there other adjacent concepts and variables to consider?
One of the earliest debates in Islamic intellectual history relates to whether legal competency (fiqh al-rāwī) is required in the transmission (riwāya) of Prophetic testimonies (ḥadīth). 19 The standard view among ḥadīth scholars is that a speaker's uprightness/probity (ʿadāla) and precision (ḍabṭ) directly impact their reliability. 20 In ḥadīth studies, the "uprightness/probity" criterion relates not just to sincerity but to other ethical and moral considerations as well. 21 Whereas the "precision" criterion relates to the memory of the speaker, their auditory capacity, and their ability to repeat things as they first heard them. Thus, according to the traditional critical method of ḥadīth scholars, uprightness/probity (ʿadāla) and precision (ḍabṭ) are necessary conditions for reliability, whereas legal competency (fiqh al-rāwī) is not. Why did precision win out over competency?
Here, the TGP is instructive. Prophetic testimonies (ḥadīth) are statements (akhbār) made by his companions about things he said, did, tacitly approved of, or descriptions of him. 22 These Prophetic statements (ḥadīth) were then passed on from generation to generation. In light of the TGP division, all of these statements (akhbār) are cases of transmission (riwāya). In other words, most hearers in the chain are having content passed on to them by a speaker whose own content is testimonially grounded. Given that most speakers in the chain are merely transmitting content they heard from those before them, legal competency seemed irrelevant to ḥadīth scholars who were instead focusing on preservation. The majoritarian position held that as long as one's auditory capacity functions properly, their memory is reliable, and they have the ability to repeat what they have heard, that suffices. 23 A direct implication of this debate is its impact on the necessary conditions for the links in the testimonial chain. Islamic scholars debated whether children ceteris paribus were reliable enough as hearers and whether they were reliable enough as speakers to partake in justified testimonial exchange. Although a more finegrained analysis is beyond the scope of this article, ultimately, Islamic scholars argued that children qua children were reliable enough as hearers to yield justification. In contrast, they held that children were not reliable enough as speakers to yield justification. 24  18 Lackey, Learning From Words, 73. 19 Al-Sakhāwī, Fatḥ al-Mughīth, 161; Al-Ṣuyūṭī, Tadrīb al-Rāwī, 162-8. 20 See, for example, Ibn al-Ṣalāḥ, ʿUlūm al-Ḥadīth, 11-2; Ibn Ḥajar al-ʿAsqalānī, Nuzhat al-Naẓar, 58. 21 The term sincerity was not used in the Islamic tradition in this context. Instead, the term uprightness or probity (ʿadāla) was used. A more thorough analysis of this condition is beyond the scope of this article. However, ultimately, the condition served a very similar purpose to the sincerity condition in the anglophone literature. 23 What is meant here is that legal competency (fiqh al-rāwī) is not required, not that precision (ḍabṭ) alone is sufficient. Of course, precision (ḍabṭ) is not sufficient for reliability. As we have mentioned earlier, ḥadīth scholars also required uprightness/probity (ʿadāla). 24 Al-Sakhāwī, Fatḥ al-Mughīth, 160-1, 305-9.
They grounded the reliability of children as hearers in the variables associated with precision (ḍabṭ), which they argued were developed enough in children above the age of seven to justify their "bearing" (taḥammul) of testimonial content. In contrast, they grounded their argument against the reliability of children as speakers on their lack of moral development. They argued that children didn't fully understand the implications of moral accountability before puberty. Thus, while children above the age of seven are generally reliable in terms of their precision (ḍabṭ), they are not reliable when it comes to their uprightness/probity (ʿadāla).
If we were to shift our analysis from a formal setting to a more natural one, the speaker criteria would likely match the hearer criteria. If children are no longer testifying to the attribution of a Prophetic statement, it would only make sense that the rigorous standards would be eased. Take, for example, an inverted form of the famous child objection case: 25 INVERTED CHILD OBJECTION 1: Mary's daughter, Kathryn, comes home from elementary school and seems a bit down. Upon asking her why she seems upset, Kathryn informs her mother that she witnessed a teacher shove one of her classmates. Mary immediately calls the school principal and demands that they do something.
Mary's response is reasonable. Despite Kathryn's young age, Mary has every reason to trust that she saw something violent occur. Even if there is some type of reasonable explanation, the fact that Kathryn witnessed the shove seems reliable enough. What if Kathryn instead transmits something she hears at school? INVERTED CHILD OBJECTION 2: Mary's daughter, Kathryn, comes home from elementary school and seems a bit down. Upon asking her why she seems upset, Kathryn whispers to her mother a few expletives which she overheard a teacher shout at one of her classmates. Mary immediately calls the school principal and demands that they do something.

Again, it seems reasonable for Mary to trust what Kathryn heard and transmitted to her. However, what if
Kathryn instead conveys to her mother the conclusion of an inference that she has made? 26 INVERTED CHILD OBJECTION 3: Mary's daughter, Kathryn, comes home from elementary school and seems a bit down. Upon asking her why she seems upset, Kathryn tells her mother that her teacher hurt one of her classmates. Mary asks Kathryn to elaborate, and Kathryn replies that she saw the teacher firmly tell her classmate to stay after class. Later at recess, Kathryn asks her classmate what happened, and instead of answering, she cries and runs back inside. At this point, Mary isn't sure what to do. This last case is fundamentally different from the first two. Although it is possible that Kathryn's judgment is correct, her conclusion is far from airtight. While Mary might be willing to accept the judgment of a reliable adult in such a circumstance, she has good cause to be reluctant when it comes to a child. If she were to call the school and demand action, clearly, the administration would need to treat this case much more gently than the previous two cases. We generally doubt the reliability of the inferential capacity of children in that age bracket. And children lack the experience to pick up on cues that are learned over the years that help us assess such ambiguous situations.
Ultimately, the main distinction between the three cases is the epistemic source of the speaker. Although many of us might not describe children as being reliable, if we consider their ability to convey things as they see or hear them, it seems much harder to disqualify them from the larger pool of adults. It is only when we are discussing more complex inferential capacities that it seems appropriate to refer to them as unreliable. 26 An objection that might be raised here is to challenge the existence of any purely sensory or testimonially-based knowledge. I do not however believe that this objection presents a serious problem for my argument because one can benefit from these distinctions in a general way without making an absolute assertion about their metaphysical make up. One could say, "pure testimony," or "pure perception," is just testimony and perception that rely on inference less than hybrid cases or cases of inference proper. Audi makes a similar point in his distinction between inferential dependency and operational dependency. Audi, "The Place of Testimony in the Fabric of Knowledge and Justification." Therefore, in order for us to distinguish between reliable and unreliable speakers, we must know the epistemic source of the content that they are communicating.

The Belief Condition
One of the core areas of disagreement in the epistemology of testimony literature has been about the belief/ knowledge condition for testimonial communication. One view, which we can call the belief-transmission view, requires that the speaker believes/knows the content that they are communicating to the hearer in order for the hearer to be able to justifiably believe or know based on the speaker's testimony. In contrast, the beliefgeneration view denies the need for any belief or knowledge condition.
There is both a strong and weak version of the belief-transmission view. The stronger version holds that every speaker, regardless of where they may fall in a testimonial chain, must believe or know the content being communicated. 27 Whereas according to the weaker view, only the original source of the testimonial chain must believe or know the content being communicated.
The TGP allows us to isolate cases where both the stronger and weaker versions of the belief-transmission view are false. In cases where a speaker's epistemic source is testimonial, otherwise known in the Islamic tradition as transmission (riwāya), it seems clear that they can serve as mere instruments in the passing on of information, preserving it from one link in the chain to another. Whether the speaker suspends judgment or believes the contrary, transmission (riwāya) allows him to pass on content to the hearer, who can then justifiably believe or come to know based on his testimony. LIGHTS OUT: Jared just moved out of his childhood apartment downtown into his first place outside the city. On his second day in his new home, while still moving his furniture, the electricity cuts out. The building manager who was helping Jared move in tells him not to worry, and that power will be restored before sunset. Jared's old place in the city would regularly get power outages, and their building manager would often reassure them that it would be restored shortly, only for that not to be the case. Due to his past experience and his unfamiliarity with the new manager, he isn't sure whether to believe that the power will be restored before sunset or not. Thus, Jared suspends judgment. A few hours later, Jared's neighbor asks him if he's heard from the manager when to expect the power to be restored. Jared informs his neighbor that the manager said that the power would be restored by sunset. Given the neighbor's long history with the manager, she comes to believe that it will indeed be restored by then.
The neighbor justifiably believes that the power will be restored before sunset based on the transmission she receives from Jared, who himself has suspended judgment on the matter. If the neighbor has reasons to believe that both the manager and her neighbor's transmission from the building manager are reliable, and they, in fact, are, it would be peculiar to deny that she is justified in believing that the power would be restored before sunset.
One objection to this line of reasoning is to deny that it is an instance of testimonial exchange due to the lack of an assertion. However, this objection can be repelled by pointing out that Jared is, in fact, making an assertion by attributing the information to the building manager. Whether the testimonial chain is made up of five or fifty transmitters, they are asserting the attribution of that content either to the original source, or at the very least, to the proximate link connecting them to the source.
According to one standard account, testimony can be formally expressed as: S testifies that p by making an act of communication a. On this account, a is a statement that is conveyed through some verbal or non-verbal form. However, in LIGHTS OUT and cases like it, something beyond the normal analysis is taking place. There are two different testimonies to consider. We could render the exchange as follows: S testifies that p by making an act of communication a, where p represents an attribution of p-2 to S-2. Thus, the first objection fails.
Another objection to this line of reasoning is to deny that it is an instance of testimonial exchange because the act of testifying is conditioned on several variables that relate back to the content of the statement, as well  27 Lackey,Learning From Words,[37][38][39][40][41] as the attitude of the speaker. As such, while LIGHTS OUT might, in fact, result in true cases of justified belief or knowledge, it would not count as a genuine instance of testimonial belief or knowledge.
This objection rests upon the assumption that testimonial knowledge is restricted to instances of testification. 28 Of course, this type of objection is guilty of begging the question. However, if one were to, for argument's sake, accept that testimony is restricted to testifying, cases like LIGHTS OUT would still need some type of theoretical and epistemological explanation. Otherwise, a portion of what we come to believe or know would be left unexplained. Given that one of the greatest impetuses behind the epistemology of testimony has been its prevalence in our lives and a simultaneous lack of corresponding theory to make sense of it, this result would be both unfortunate and undesirable. Furthermore, it would leave these objectors with a diminished theory of knowledge.
A likely conclusion to this response would be the invention of a new term to describe this non-testimonial source of knowledge. In that case, a term such as transmission (riwāya) should be suitable. However, if this is the real nature of the dispute, then the disagreement is dangerously close to being verbal. The second objection has resulted in an acceptance of LIGHTS OUT as a case of justified belief or knowledge through transmission (riwāya) while insisting that it be referred to as something other than testimony.
The objection is still substantive in that it entails that sensory testimony (shahāda) and transmission (riwāya) are not two species of some larger set. But it is not entirely clear why these two acts of communication, in which content p is conveyed to the hearer, are not two forms of the same act. The rationale seems to be that the act of sensory testimony (shahāda) and the act of transmission (riwāya) are so fundamentally different that the relation between them is disjoint. This, however, misses the point; one can easily concede that the two acts are disjoint while maintaining that they are both strictly included in the larger set of acts of communication.
Take, for example, Lackey's disjunctive definition of testimony: DVNT: S testifies that p by making an act of communication a if and only if (in part) in virtue of a's communicable content, (I) S reasonably intends to convey the information that p or (2) a is reasonably taken as conveying the information that p. 29 It is quite clear that formal acts of sensory testimony (shahāda) are included in this definition. The real test is whether LIGHTS OUT can also be included in either of the disjuncts. By moving away from characterizing the epistemic source of testimony as a particular speech act, Lackey grounds the definition in both the intention to communicate and in the conveyance of information. Thus, Jared only needs to intend to communicate with his neighbor what the building manager said, or his neighbor needs to take his utterance as reasonably conveying its propositional content. The case of LIGHTS OUT suffices both sides of the disjunct. Thus, the TGP has helped us disprove the stronger thesis through the counterexample of LIGHTS OUT. However, what about the weaker thesis? Those who hold the weaker thesis claim that it is only the first speaker in the testimonial chain who must believe or know the content of the testimony. 30 If the first speaker in the chain does not believe the content in question, it is posited that the hearer further downstream will not have properly justified reasons to believe the content being transmitted.
While proponents of the weaker thesis are on to something epistemically significant, they've ignored the epistemic source of the speaker and have thus misdiagnosed the problem. I submit that cases where a lack of knowledge or justified belief on the part of the speaker prohibit justifiable testimonial exchange are due to the epistemic source upstream and not the absence of belief on the part of the speaker. The real issue is not whether the speaker has justified belief in the proposition asserted but what type of source the original testimony is epistemically grounded in.
 28 By testification, I mean any view that is going to require some type of intentional attitude on the part of the speaker in order for their act of communication to be considered testimonial. Those who advocate for the belief-transmission view are, by definition, associating testimony with testification. 29 Lackey, Learning From Words, 35-6. 30 Ibid., 41.
Lackey's CREATIONIST TEACHER has already made the case that the belief of the testifier need not follow the evidence for it to yield testimonial knowledge. 31 The TGP not only allows us to better understand the case of CREATIONIST TEACHER, but it also allows us to propose counterexamples to the belief-transmission view where the testifier is following the evidence and yet testifies to the contrary of their belief. Take, for example, the case of DETROIT ONCOLOGIST: DETROIT ONCOLOGIST: Ahmad is an oncologist who works at a cancer center in the Metro Detroit area. A young couple comes into his office to ask about a new miracle cancer drug for their child that has just been granted FDA approval. Despite the long debate over the efficacy of the drug, ultimately, the FDA found the evidence convincing and has given the drug manufacturers the green light. Ahmad, having specialized in cutting-edge cancer research, is not just familiar with the range of scholarly opinion on the topic but with the studies and the rationale for and against the miracle drug. Given Ahmad's interpretation of the findings of the studies undertaken before FDA approval, he is not convinced that the drug works. He believes the FDA rushed to judgment, and while the drug might not be dangerous, it is not useful either. However, he also recognizes that many other oncologists, researchers, and the FDA disagree. Furthermore, he is sensitive to the fact that cancer patients and their families are often left with few options. Thus, he tells the couple that the drug has been shown to be effective and could help cure various types of cancer. Based upon his answer, the couple not only comes to believe that the drug works, but they also go on to rely on it as a primary means of treatment for their child.
Is the couple justified in believing that this new drug has been shown to cure cancer? According to the weaker belief-transmission view, the answer is no. However, this doesn't seem right.
Ahmad is the original testifier in this case; he is not transmitting someone else's testimony since he, himself, has direct access to the evidence for and against the new drug. The relationship between Ahmad's testimony and the truth of the matter is not accidental. Ahmad is aware of the rationale of the other physicians and the alternative interpretations of the studies of the drug, which hold that it is indeed effective. Furthermore, Ahmad recognizes that experts can disagree over the interpretation of evidence. Nevertheless, he is not convinced of the efficacy of the drug and does not believe it can cure cancer.
He communicates to the couple that the drug can, in fact, cure cancer based upon these alternative interpretations of the findings, all the while disagreeing with their rationale. The fact that Ahmad calls to mind these other interpretations of the studies undertaken means that the connection between the couple and the truth of the matter is not severed. If it turns out that the miracle drug works, the couple knew this to be the case given their rational trust in Ahmad and given Ahmad's reason-based conclusion. The result is that the rationale Ahmad is basing his testimony on comes apart from his belief in the content. Therefore, the couple does, in fact, justifiably believe that this miracle drug cures cancer, even though their source, Ahmad, did not believe it to be the case.
How can the TGP help us make sense of DETROIT ONCOLOGIST? Does the initial speaker's epistemic source help us make sense of why DETROIT ONCOLOGIST or other similar cases can serve as counterexamples to the necessity thesis concerning belief? DETROIT ONCOLOGIST and similar cases involve a substantial level of inference on the part of the initial speaker, otherwise known in the Islamic tradition as opinion (raʾy). Given that this inferential process allows for rational disagreement and given that the testifier is aware of an alternative rational interpretation of the findings, the tether between the initial speaker and the truth of the matter is never fully broken. Whether or not they are convinced by that rationale becomes irrelevant. In such cases, our reason allows us to "hold" more than one set of arguments in mind, even if we take one to be true and the rest to be false. In contrast, knowledge gained from our sense perception does not allow for the same type of rational disagreement and multilayered reasoning. The TGP has thus helped us disprove both the stronger and weaker versions of the belief-transmission view.

Conclusion
Testimony, as a source of knowledge, is by its very nature social and anti-individualistic. Given this fact, a hearer's knowledge and justification cannot be properly evaluated without considering the reliability of the speaker. The reliability of the speaker is, in turn, at least partially grounded in their epistemic source. The epistemic source of their testimony can be grounded in sense perception, inference, or further testimonial exchange. Understanding this distinction helps us better understand the nature of testimony as a source of knowledge. More importantly, it helps us evaluate the reliability of testimony in our day-to-day lives.

Conflict of interest:
Author has no conflict of interest.