Month: May 2016

I. Introduction

 

1.1 In this chapter, Pollock and Cruz (1999) argue for direct reliablism as a viable answer to, “what are the actual norms governing human epistemic competence?” (191). Their argument for direct reliabilsm stems from their rejection of the doxastic assumption and claim that epistemic norms coincide with nondoxastic internalism. P&C realized that holistic coherence theories fail because they cannot differentiate between justified and justifiable beliefs. They concluded that reasons play an essential role in justification.

P&C critique foundationalism and doxastic theories because they cannot accommodate perceptual knowledge and or describe human rational cognition. In vision, beliefs about physical objects can be derived from the percept either (A) directly from the percept or (B) inferred from beliefs produced by the percept that describes the percept. Unlike foundationalism, which views (A) impossible, P&C argue that such epistemic norms are possible, yet unnecessary. The fundamental principle of direct realism is that inferences and beliefs are derived directly from the percept. Therefore, direct realism is similar to foundationalism, except the foundations are percepts and not beliefs.

 

1.2 levels of epistemological theorizing

 

In this section, P&C explain three levels of epistemological theorizing. In the low-level, philosophers participate in bottom-up theorizing by investigating kinds of knowledge claims. The intermediate-level involves the investigation of topics that pertain to all kinds of knowledge claims explored on the lowest-level. The highest-level is considered top-down epistemological theorizing because it harbors general epistemological theories that try to describe “how knowledge in general is possible” (192). P&C claim that epistemologically theorizing requires both bottom-up and top-down processes. Specifically, one must first use top-down theorizing to argue for a high-level theory and then form conformable low-level theories to support it. If one cannot find such low-level theories, the high-level theory should be abandoned. In Section II, P&C argue that defeasible reasoning “provides the inferential machinery upon which to build low-level theories of epistemic norms governing specific kinds of knowledge” (200).

 

1.3 filling out direct realism

 

P&C argue for the high-level theory of direct realism by constructing compatible low-level theories. Construction is defined as “describing the various species of reasoning that can lead to justified beliefs about different subject matter” (195). This is the main goal of the OSCAR project. The creation of an artilect depends on successful and detailed construction of low-level depiction of our epistemic norms so that a computer system can encode such norms (194). P&C are open to epistemologists that disagree with direct realism, but they doubt whether an artilect can be created on opposing theories of epistemic foundations.

II. Reasoning

 

Direct realism requires epistemic norms that can appeal to perceptual states but not necessarily our beliefs about that state. P&C state that there can be “half-doxastic connections” between beliefs and non-doxastic states that is analogous to the structure of ordinary defeasible reasons. The only difference in reasoning is the reason-for relation because different states with similar content can support different inferences.

 

P&C state the definition of reason as:

 

A state M of a person S is a reason for S to believe Q if and only if it is logically possible for S to become justified in believing Q by believing it on the basis of being in the state M.

In other words, the state M does not need to be a belief. The fact the ball looks red to Bob (P) is enough reason for Bob to believe that it is red (Q).

2.2 Defeaters

Defeaters for half-doxastic connections operate like defeaters proposed in the foundations theory. It is important to characterize defeaters for defeasible reasons in low-level accounts (201). P&C describes two kinds of defeaters. The second type of defeater is redefined to include nondoxastic states.

(1)     REBUTTING DEFEATER: If M is a defeasible reason for S to believe Q, M* is a rebutting defeater for this reason if and only if M* is a defeater (for M as a reason for S to believe Q) and M* is a reason for S to believe ~Q.

(2)      UNDERCUTTING DEFEATR: If M is a nondoxastic state that is a defeasible reason for S to believe Q, M* is an undercutting defeater for this reason if and only if M* is a defeater (for M as a reason for S to believe Q) and M* is a reason for S to doubt or deny that he or she would not be in state M unless Q were true.

A rebutting defeater (1) is a reason that denies the conclusion (Q). For example, if Bob colorblind and believes that his colorblindness is such that whenever something looks red it is actually green, he now has reason to believe the ball is not red. An undercutting defeater (2) is a reason that causes a person to no longer believe Q without negating it. This defeater attacks the connection between evidence for the reason and the conclusion by showing that one’s reason for believing Q doesn’t mean that Q is true. In this example, Q is the ball is red. If Bob is informed that the ball is being irradiated by red lights, Bob no longer has reason to believe that P, the ball appearing red, is actually red, rather than having a reason for saying that it is not red. Undercutting defeaters are reasons for “P does not guarantee Q,” which is abbreviated (P ⊗ Q) (197). In other words, the irradiation means that Bob’s belief that it is red doesn’t guarantee that it is red.

2.3 Justified Beliefs

In direct realism, beliefs are justified by reasoning (197). P&C define reasoning is as constructing longer arguments out of shorter ones, or subsidiary arguments. Each argument is a sequence of beliefs and nondoxastic mental states that are ordered such that each member is either (1) a nondoxastic mental state or (2) there is a proposition(s) or nondoxastic state earlier in the sequence that is reason for P (197). An argument is instantiated if a person is in a nondoxastic state and believes each proposition on the basis of earlier propositions.

Inference-graphs are a set of arguments that shows the construction of arguments. Each node is given a status- assignment, which assigns what inferences are defeated or undefeated. A partial-status assignment assigns defeated or undefeated to subset with the following rules:

1. if A is a one-line argument (i.e., a single percept), A is assigned 
“undefeated”;
2. if some defeater for A is assigned “undefeated”, or some member 
of the basis of A is assigned “defeated”, A is assigned “defeated”;
3. if all defeaters for A are assigned “defeated” and all members of the basis of A are assigned “undefeated”, A is assigned 
“undefeated”.

In other words, an argument A is undefeated if and only if every argument in the graph is assigned undefeated to A.

Figure 7.1 illustrates the importance of defeasibility in justification of arguments. The arrows represent an inference from each node. Both P1 and Q1 are nondoxastic states.

 

Because the conclusion of the second argument is an undercutting defeater, Bob is not justified in believing P3 and the first argument is assigned defeated. (P2 ⊗ P3) is a defeater for an argument because it supports a defeater for a final step. However, if Bob finds a defeater for some part of the second argument, the first argument is still possible. An argument can be defeated if it is (1) based on defeated subsidiary argument, or (2) has undefeated defeater. Therefore, arguments are “provisional vehicles of justification” because arguments can defeat each other and that a defeated argument can be reinstated.

Figure 7.2 illustrates the concept of collective defeat.

Collective defeat is the situation where two or more arguments defeat each other. In this example, we have good reasons for believing that it is and isn’t raining. Since both conclusions are defeated by one of two possible status-assignments, both arguments are defeated relative to the inference graph. Therefore, we should not accept either conclusion.

(1) We assign undefeated to P1, P2, “Jones says it is raining,” “Smith says that it is not raining”, and “It is raining”, and defeated to “Smith says it is not raining.” Conclusion: it is raining.

(2) We assign undefeated to P1, P2, “Jones says that it is raining” “Smith says that it is not raining”, and “It is not raining”, and assigns “defeated” to “It is raining”.

An argument is provisionally defeated if a status assignments assigns defeated to it and another status assignment assigned undefeated to it. Unlike an argument that is defeated outright, it can still defeat other arguments.

 

Figure 7.3 illustrates provisional defeat

 

Smith and Jones accuse each other of lying. One status assignment assigns defeated to “Smith is a liar” and the undefeated to “Jones is a liar” and the other assignment does the reverse. If Smith is a liar, then the inference that it is raining from Smith is defeated. However, even if “Jones is a liar” is undefeated and “Smith is a liar” is defeated, the inference saying “it is raining” is still defeated. Therefore, this argument is provisionally defeated because the inference “Smith is a liar” is defeated and undefeated by separate status assignments and can defeat the inference that it is raining.

 

III. Perception

 

P&C claim that direct realism can solve the problem of perception, or how we can gain knowledge of the external world through perception. P&C consider the ability of perception to provide reasons for judgment about the world as the fundamental principle of direct realism (201). Section I states that the inference is produced directly from the percept, not indirectly through beliefs about the percept.

 

PERCEPTION:

Having a percept at time t with the content P is a defeasible reason for the cognizer to believe P-at-t.

 

P-at-t is the percept a reasoner believes at time t. P&C claim that this principle is the most basic component of rational cognition; it cannot be justified. This principle must be present because it is “an essential ingredient of the rational architecture of any rational agent” (201). Reliability defeaters are undercutting defeaters for perception. They prove that an inference from a percept is unreliable under present circumstances.

 

Perceptual-reliability:

Where R is projectible, “R-at-t, and the probability is low of P’s being true given R and that I have a percept with content P” is an undercutting defeater for PERCEPTION.

 

P&C call for importance of the projectibility constraint in reliability because it involves whether a percept is retained over time. Consider the example:

 

You have two circumstances. C1: Bob was born in 1998. C2: Bob is wearing rose-colored glasses. If Bob is wearing the glasses, C2 is a reliability defeater and it is unlikely that the ball that appears red to Bob is actually red. However, if Bob was born in 1998, this enables the possibility that Bob could also be wearing rose glasses. In the disjunctive circumstance (C1 v C2), there is a high probability that Bob could be wearing rose glasses. Therefore, it is also unlikely that the ball is actually red. This example shows that disjunctive circumstances present an indirect defeater to perception.

 

IV. Implementation

 

This section illustrates how reason schemas are implemented into OSCAR. We will clarify a few terms involved in implementation. OSCAR reasoning is the construction of both deductive inference rules and defeasible reason-schemas (203). Inputting the premises generates the queries, or “epistemic interests.” Reasoning through queries lead to conclusions and that are computed as inference graphs.

OSCAR performs bidirectional reasoning: the agent reasons forward from the premises (forward-reasons) and backward from the queries (backward-reasons) (203). Simple forward-reasons have no backward-premises and simple backward-reasons have no forward-premises. Mixed reasons contain backward- and forward- reasons. In simple reasons, the conclusions can be directly inferred from the reasons. In contrast, conclusions in mixed reasons are made only if (1) the reasoner adopts interest in the backward premises, and (2) those interests are discharged. Interest in backward premises is adopted only when inference nodes are constructed that present forward premises, and vice versa. The bidirectional use of each type of premise on the other provides control over reasoning progression.

 

The problem lies in how to implement perpetual reliability. The definition of perceptual-reliability is adjusted to consider reason strengths:

Perceptual-reliability:

Where R is projectible, r is the strength of PERCEPTION, and s < 0.5 ⋅(r + 1), “R-at-t, and the probability is less than or equal to s of P’s being true given R and that I have a percept with content P” is an undercutting defeater for PERCEPTION.

 

Reason strengths range from (0-1) but are mapped to probabilities in the interval (0.5-1) (206). P&C first propose this as a backward-reason, but subsequently claim that because that there are no constraints on R, the reasoner would spend too much time attempting to determine reliability given everything in the situation. P&C instead propose it as a degenerate backward-reason, with no backward premises, and P-at-t and the probability premise as a forward-premise. How can we implement perceptual reliability if we need to know if R is true at the time of the percept, but we can only infer it from the fact that R was true earlier?

 

V. Temporal Projection

 

This section of the chapter opens with a discussion of the strengths of PERCEPTION while acknowledging its major shortcoming: perception, at best, is nothing more than a “form of sampling.” That is, it is not possible for a cognizer to continually perceive and process the state of everything in his or her surrounding environment. Rather, individuals perceive small “space-time chunks” of their environments and make perceptual inferences about the state of the world at large by combining these chunks. The problem with this process of forming inferences, P & C argue, is that there is a surprising difficulty in drawing accurate conclusions about the world at large based on combinations of perceptual samples.

 

A large part of this difficulty involves the lack of time-sensitive stability exhibited by the majority of objects in the natural world. Making inferences based on single perceptual samples of given objects presupposes that the properties observed are stable over time, which is often not the case. Theoretically, an individual would need to observe the same object at multiple points in time in order to determine whether or not its properties had changed; only when affirming that they had remained unchanged could its stability be inferred, and broader inferences about its nature be made. However, making observations of the same object at various times requires the observer to accurately reidentify the object, a task that can become impossible when the object at hand rapidly or unpredictably changes its properties.

Thus, an assumption of some stability must be made about objects that an agent uses in the process of forming inferences about the world. P & C argue that an object is considered stable if, given that it is observed to hold true at an initial point in time, the probability is high that it will continue to hold true at a later point in time. They furthermore argue that the probability that the property of a given object will continue to hold true over time decreases as a function of the length of the time interval. P & C call the process of assessing the stability of an object based on the consistency of its properties over time temporal projection. Temporal projection, they argue, is essential in the rational assessment of property stability, and thus in the process of forming inferential conclusions about one’s surrounding environment. What does temporal projection look like when applied? In other words, how should temporal projection be implemented?

VI. Implementing Temporal Projection

 

In this section, P & C discuss a great deal of complex algorithms, atomic formulas and codes, which will not be discussed in detail here. Without discussing these formulas in great detail, it will suffice to consider P & C’s explanation of temporal projection as dependent on temporal projectibility. Intuitively, in order to temporally project the stability of a given object’s property of time, the properties of that object must be temporally projectable; i.e. it must be possible for a rational agent to determine their constancy based on probability. The remainder of this section is an analysis of atomic formulas and algorithms used in the process of temporal projection, all of which P & C argue are, in fact, temporally projectable.

 

VII. Reasoning About Change

 

In a vein similar to that of Section V, this section discusses the need of a cognizer to account for the tendency of most, if not all, objects in the natural world to change as a function of time and other variables. Moreover, P & C address the need of a rational agent to consider this tendency to change when making broader inferences about his or her surrounding environment. In their discussion of this need to account for change, P & C identify four kinds of reasoning:

1. First, they argue that the agent must be capable of acquiring perceptual information about the surrounding world. This implies a necessity of proper cognitive functioning and reliably sensory interactions.
2. Second, the agent must be able to combine isolated perceptual chunks of his or her surrounding environment into a coherent picture of the broader world.
3. Third, the agent must be capable of perceptually detecting changes in previously identified components of his or her broader to picture of the world, and to amend this picture accordingly.
4. Lastly, the agent must be capable of acquiring causal information about “how the world works” and to use this information to efficiently predict patterns of change that may result in the future, either from uncontrollable, natural circumstances or from the agent’s own actions.

The remainder of this section discusses the fourth type of reasoning in-depth, mentioning that the ability to foresee change necessitates the ability to foresee non-change; P & C write on page 219 that “…reasoning about what will change if an action is performed or some other event occurs generally presupposes knowing what will not change.” The remainder of the section is more or less an elaboration upon the logic of this claim, with a focus on the argument that predicting what will likely occur is largely dependent on knowledge about is not likely or impossible to occur.

 

 

VIII. The Statistical Syllogism

 

            In this section, P & C continue to build on their foundational claim that an individual’s ability to rationally navigate through the world depends heavily on his or her ability to make reasonable predictions about changes that may take place in his or her environment under various circumstances. They argue that, in order to function in a complex environment, such as the natural world as we know it, a rational agent must be equipped with rules that:

• enable the agent to form beliefs in statistical generalizations, and;
• enable the agent to make inferences based on those statistical generalizations that are applicable to individual circumstances. (pp. 229-230)

P & C provide an archetypal, non-numerical version of the statistical syllogism in their “most Fs are Gs” example:

 

Most F’s are G’s

This is an F

_____________

This is a G.

 

P & C explain that, because human beings often reason this way, a rational execution of such logic is essential in making reasonable predictions about the state of one’s surrounding environment. The remainder of this section involves a series of statistical and algorithmic examples supporting the validity and applicability of this claim.

 

IX. Induction

            This section relies heavily on the claims made in section VIII about the need for a rational agent to effectively make reasonable predictions about the world based on generalizations; these generalizations, they argue, can be either exceptionless (all Fs are Gs) or statistical, varying in probability (the probability of A being B is high). We become justified in believing the statistical generalizations we make through a process of induction. 

P & C begin with an explanation of the simplest kind of reasoning, enumerative induction, which involves a process of generalization based on sampling (i.e. “all As in sample X are Bs, so all As are likely Bs”). The most important defeater to consider when evaluating this line of reasoning is the possibility that X is not a reasonably “fair” or accurate sample; that is, it does not accurately encapsulate or characterize the population that it supposedly represents. The “fairness” or accuracy of a sample, i.e. its reliability in the formation of conclusions about a represented population, depends on a number of factors, including sample size and sample diversity.

P & C argue that a second kind of induction, termed statistical induction, is much more important for rational agents in their process of forming conclusions about the world based on observation of samples. P & C succinctly summarize the principles of statistical induction in 7.6:

“If B is projectible with respect to A, then ‘X is a sample of n A’s r of which are B’s’ is a defeasible reason for ‘prob(B/A) is ap- 185. Fair sample defeaters are discussed at greater length in Pollock (1989), but no general account is given. 186. Further discussion of this can be found in Pollock (1989, 315ff). proximately equal to r/n’”

The remainder of the article involves the analysis of a number of algorithmic expressions and likelihood ratios that illustrate the process of statistical induction. One might assume that, like temporal projection, the justification of an inductive argument depends on a number of factors such as the constancy of observed properties over time, representativeness of the sample, etc. However, P & C conclude this section by arguing that the strength of induction is in its lack of need for justification (pp. 237). They add, however, that “…principles of induction are philosophically trouble free. Major questions remain about the precise form of the epistemic norms governing induction. Suggestions have been made here about some of the details of those epistemic norms, but we do not yet have a complete account.” (p.237).

 

Discussion Questions

1) P&C present that the task at hand is to construct low-level theories that support direct realism (192). How does bias fit into this claim? Specifically, are the low level theories constructed only to match with it, instead of accurately construct reality?

2) Does direct realism really avoid the problems of justification that foundationalism runs into?

3) The original example for statistical syllogism is “no one believes everything in the newspaper to be true – but I do believe that most is true and that justifies me in believing individual newspaper reports (230)” Can justification be based on the “most is true” concept?

4) In their discussion of perception in the role of forming inferences about the world, P&C lend a great deal of weight to the agent’s ability to integrate numerous perceptual “space-time chunks” into a coherent, broader view of one’s surroundings. Can you give any real-world examples of this process?

5) P & C claim that the strength of induction lies in its lack of need for justification. Do you agree with this? Is induction an infallible process?

 

Peer Commentary:

Arguing, reasoning, and the interpersonal (cultural) functions of human consciousness”

Baumeister, Masicampo, and DeWall agree with M&S that reasoning is an interpersonal rather than individual exercise, and that reasoning was evolutionarily advantageous for humans. The purpose for reasoning, they posit, was to advance culture – individuals that could reason well (and thus propagate culture by creating interpersonal connections) were selected for. B, M, & D also argue that thinking and consciousness exists in order for us to share it with others – “much of thinking is for talking” (74).

Regret and justification as a link from argumentation to consequentialism

Connolly & Reb support M&S’s ideas on the evolutionary role of argumentation, but believe that emotions can be a critical link between argument making and consequential decision making. The phenomena of regret, regret avoidance, and justification are not obstacles to good reasoning but can actually facilitate decision making. The attraction effect is the phenomenon in which an option is seen as more attractive when compared to a separate, irrelevant option (75). External accountability can exacerbate the attraction effect, while regret priming – demanding to justify one’s decision to oneself – can eliminate this effect as the goal is to arrive at a conclusion that aligns with one’s values rather than just persuade others.

The freak in all of us: logical truth seeking without argumentation

De Neys attacks M&S’s presentation of peoples’ performance on classical, non-argumentative logical tasks. De Neys concedes that people do perform poorly in non-argumentative tasks, but people reason better in argumentative contexts doesn’t mean that they don’t try to reason logically outside of this context. Citing a number of psychology and neuroscience studies, De Neys presents data showing better reasoning by people in the classical reasoning tasks. He argues that even thought they are wrong, there is at least evidence that they are trying to be logical. In pointing this out, De Neys argues that classical reasoning tasks are less artificial than M&S suggest (76).

Reasoning as a lie detection device

Dessalles argues against the biological function of argumentation presented by M&S. M&S argue that the optimization of communication derived from argumentation occurs at a group level, but Dessalles points out that evolution works at the individual level rather than the group level. Reasoning, then, is for lie detection and aims to restore consistency to communication – remarkably similar to M&S’s proposed function of reasoning. According to Dessalles, the social benefit to the individual of exposing inconsistencies and restoring consistency allows the individual to compete well with their peers. There are side effects to this model of reasoning for social benefit, including invoking evidence that cannot be verified.

Reasoning is for thinking, not just arguing

Evans argues that while reasoning may have evolved primarily for argumentation, that it is now used for other functions. He states that M&S have been dismissive and limited when addressing the dual process theories, overlooking the importance of general, heritable intelligence in solving novel reasoning and decision making. To support this Evans provides evidence that the ability to solve novel problems is related to intelligence and the ability to reason, contrasting M&S’s argument that intuition is better. Evans disagrees with M&S’s ideas about the evolution of reasoning, pointing out that the evolution of higher cognitive abilities was not driven by Darwinian pressures, otherwise other animals would have evolved them as well.

Artificial cognitive systems: where does argumentation fit?

Fox brings findings from research with artificial intelligence systems into the discussion of the mechanisms and functions of argumentation in social interactions. He argues that reflective reasoning has more than just social benefits for the cognitive agent (human or artificial), and uses a formal model to further clarify M&S’s distinction between intuition and reasoning. He also argue that the argumentative theory is what is meant by evidence by determining what kinds of statements can and cannot be used in argumentation.

Reasoning, argumentation, and cognition

Frankish agrees with M&S that arguments enhance communication, but that this is perhaps primarily for enhancing collective cognition and that other social functions of arguing exist such as finding a mate. Reasoning, he argues, may have evolved for public use, but has been co-opted to serve individual cognition, in which there are more than just epistemic motives at play but also social motives. Individualized reasoning is just an internalized version of public argument that precedes individual reasoning. M&S’s theory of personal-level reasoning, Frankish argues, requires intuition regarding the rules of inference which can be either abstract (requires explicit learning) or linguistic (learned by exposure to arguments of others).

Reasoning as a deliberative function but dialogic in structure and origin.

Godfrey-Smith and Yegnashankaran argue that reasoning is an internalized form of interpersonal exchange of ideas, and has several functions but is primarily for deliberation. As a persuasive tool, reasoning in the form of dialogue is more adept at reaching conclusions and justifications. G-S & Y also find tension in M&S’s claims that people are poor individual reasoners but function well in groups, but concede that M&S’s theory better explains the existence of confirmation bias.

 

Understand, evaluating, and producing arguments: training is necessary for reasoning skills

Harrell refutes the claim by M&S that people are good arguers by citing experimental evidence to the contrary, including some of the same studies cited by M&S. She argues that the evidence presented by M&S only vaguely defines argumentation skills and provides poor evidence that people can understand, evaluate, and produce arguments. The literature, according to Harrell, actually shows that these skills are poor in untrained people, but that following formal argumentation training performance improves.

The argumentative theory of reasoning applies to scientists and philosophers, too
Johnson explores what the implications of M&S’s theory of argumentation are on the roles of professional reasoners such as scientists and philosophers. If the function of reasoning is the same for every person in society, as M&S implicitly state, does this mean that scientists and philosophers are in the business of persuasive argumentation – convincing people of their ideas – rather than seeking the truth? This must include M&S themselves as well in order to avoid the non-reflexive fallacy. M&S challenge the idea of scientists and philosophers as “elite thinkers”, so does every human have this capacity? Additionally, does this mean that scientists and philosophers are only governed by confirmation biases like the rest of us? “Research does not begin dispassionately” (81-82), and Johnson believes they should accept this as a reality of their profession.

True to the power of one? Cognition, argument, and reasoning
Khlentzos and Stevenson support the argumentative theory of reasoning presented by M&S, but question the way M&S have divided functions between systems 1 and 2, specifically how system 2 is a “backup” for system 1. K&S propose an alternative role for S2, as a reasoner that filters S1 outputs and independently produces conclusions that are subject to revision given new evidence. They concede that S2 plays some kind of regulatory role because S1 is both probabilistic and deductive. K&S disagree with M&S that confirmation bias is not a flaw and state that it can polarize conversation and dissuade argumentation.

What people may do versus can do
Kuhn seeks to expand on the power of human reasoning supported by the argumentative theory. She states that while can reason fairly well under ordinary conditions, numerous studies involved in the training of adolescents has shown that in supportive environments, people can reason can show much stronger argument skills. This supportive environment is “sustained engagement of adolescents in dialogic argumentation” (83), akin to a longitudinal study rather than the more cross-sectional studies cited by M&S. With this sustained training, over a few months argumentation improves greatly in participants. Kuhn is hopeful about applications of this and the possibility for universal improvement of argumentation with formalized training programs, and the establishment of a new norm of argumentative skills.

The need for a broad and developmental study of reasoning
Narvaez argues that M&S’ by focusing purely on a rhetorical use, they overlook the practical uses of reasoning, and therefore have too narrow of a view of the uses of reasoning. She notes that M&S overlook sociopolitical reasoning used to design laws, everyday reasoning used to take appropriate courses of action, and goal-motivated reasoning used to plan and reflect on failure and success. Her other main point takes issues with most of M&S’ research findings supplied by college students, noting that older adult subjects may have developed better inductive reasoning, while also pointing out the issue in generalizing human nature from the Western, Educated, Industrialized, Rich and Democratic (WEIRD) population.

Putting reasoning and judgment in their proper argumentative place

Oaksford generally argues with the thesis provided by M&S, but would like to see a more probabilistic analysis when judging the strength of an argument. They use the denying the antecedent fallacy to provide an example where the change in the degree of belief brought by an argument is useful in the judgment of the argument.

On the design and function of rational arguments

Opfer and Sloutsky propose three obstacles to reasoning as an argumentative tool. The first is that changing beliefs and attitudes would not be affected by solid reasoned argumentation, citing studies that show those that are less confident in their beliefs often yield to their more confident peers. The second is that emotionally charged examples often persuade more than reason, which takes up more cognitive resources. The final is their distinction between linguistic and argumentative “operators” and “receivers.” They contend that a more proficient language  user (operator) can aid the lesser (receiver) to a better use of language, but not in the case of more and less reasoned argumenters.

What is argument for? An adaptationist approach to argument and debate

Pietraszewski agrees with M&S’ thesis but wishes to answer what argument serves a purpose for if it is so that reasoning serves argument. They stray away from the classical conclusion of seeking truth and accuracy and instead propose that argument and communication exists to affect behavior, which will then alter the future actions of others in regards to the communication. This leads to 2 classes of argument psych: to deal with conflicts of interest and to socially coordinate. They follow the second class to be the evaluation of argument as “who is arguing should be just as important as what they are saying when considering the ‘goodness’ of the argument” (87).

The importance of utilities in theories of reasoning

Poletiek proposes that M&S’ theory does not properly address different reasoning contexts outside of an argumentative one. She uses the example of a hypothesis testing experiment where participants came up with different outcomes when in an argumentative motive versus other motives (other than determining the truth) to highlight the need of other utilities, such as signal-detection theory, to reason in a variety of contexts.

When Reasoning is persuasive but wrong

Sternberg takes issue with M&S’ claim that reasoning could have evolved out of a function to argue. They use an example 2 individuals arguing of the existence of a threat to, independent of who was correct, their reasoning was purely argumentative, but the correct individual survived. Put in the context of global warming, Sternberg contends that this model of reasoning will doom human survival and that reasoning should function to serve “veridicality” (89).

The chronometrics of confirmation bias: Evidence for the inhibition of intuitive judgments

Stupple and Ball use chronometric data of participants in reasoning tasks to disprove this idea brought forth by M&S that people first try to reason to serve their existing beliefs. They found participants spent most time reasoning through a believable but invalid proposition, resisting their intuitive judgments. With this evidence they conclude participants actually could reason in search of the truth while inhibiting their intuitions.

Spontaneous inferences provide intuitive beliefs on which reasoning proper depends

Uleman, Kressel, and Rim take issue with the over simplification of the M&S’ intuitive beliefs and provide evidence for the spontaneous inferences existing as what M&S think to be intuitive beliefs. They provide evidence for these spontaneous, unconscious inferences leading to the formation of conscious reasoning with an experiment in which participants familiarized themselves with the sentence “John returned the wallet with all the money in it” and in other tasks always associated John with honesty (90).

Query Theory: Knowing what we want by arguing with ourselves

Weber and Johnson see query theory oversimplified in M&S’ article as only an example of reason based choice, where they see it as evidence for the retrieval of implicit memories to evaluate choices leading to decisions and implement into arguments. Ultimately the processes shown by QT imply that argumentation is not only interpersonal but also leads to intrapersonal implicit preference construction. Both seem to have implications in argumentative contexts.

Reasoning, robots, and navigation: Dual roles for deductive and abductive reasoning

Wiles explores another aspect of cognition in regards to reasoning and argues for the existence of a more primitive function shared with other animals and modeled in robots and mice which is navigation. They find systems that navigate well to be abductive reasoners adding another aspect of reasoning not accounted for when viewing reasoning only as argumentative.

Some empirical qualifications to the arguments for an argumentative theory

Wolfe sees that data not accounted for by M&S actually indicate people aren’t actually as good at evaluating arguments at M&S say they are, citing several studies that went overlooked by M&S. They then point out a very key distinction not made by M&S: “confirmation bias typically refers to a biased search for or weighing of evidence, whereas myside bias refers to biases in generating reasons or arguments” (93). They have shown myside biases can actually be reduced with training. They ultimately propose that we actually do have the resources to form reasoned arguments on our own otherwise argumentation would not work.

Deliberative democracy and epistemic humility

Chien-Chang Wu wishes to apply M&S’ ideas into a political framework called deliberative democracy in order to apply group reasoning into public policy decision. He notes that M&S’ theory of reasoning fits well with deliberative democracy and focuses on an epistemic aspect, pointing out that there are complicated issues embedded in applying M&S’ theory that they don’t address. He lays out three conditions that he sees as necessary in order for deliberative democracy to produce epistemic good: considerations of ethics, a “deflationist” definition of truth (noting realist truth unlikely achievable), and considerations of the framing powers.

 

Author’s Response. Argumentation: Its adaptiveness and efficacy

R1. Different Definitions of Reasoning

M&S explain how reasoning, as they describe it, is a form of higher-order intuitive inference with a specialized domain and task, which contrasts with ordinary intuitive inference. Some commentaries defend a different definition of reasoning. Khlentzos & Stevenson suggest that some type of system 2 reasoning must have evolved to arbitrate between contradictory system 1 outputs (for instance when perception contradicts memory). They argue that reasoning is specifically geared toward this end, which M&S argee would be true with a much broader definition of reasoning. Poletiek and Narvaez both argue that reasoning guides strategy and action choice, which M&S argue is a function of intuitions and falls outside the scope of their definition of reasoning. A few commentaries raise additional mechanisms in System 2 (hypothetical thinking, elaborative planning, and avoiding decisions we would regret) that they argue directly lead to good outcomes without involving argumentation. M&S again argue that these mechanisms do not qualify as reasoning under their definition. They do think it would be interesting for researchers to consider System 2 as comprising several different mechanisms other than reasoning because it could explain the covariation of traits measured by various measures of cognitive ability. M&S end this section by saying that offering another reasonable and useful definition of reasoning is not enough to object to their definition.

R2. Evolution and function of reasoning

M&S argue that a number of objections were based on a misunderstanding of their hypothesis on the evolution and function of reasoning. They assert that they never argued that reasoning is designed only to find arguments to persuade others, or that epistemic goals should be poorly served by reasoning, or that mere rhetoric is all it takes to influence people, or that people hardly ever change their mind, as many authors believed. M&S apologize for devoting more space in their article to the production of arguments by communicators (rhetoric) than to the evaluation of arguments by the audience (epistemic). They say the argumentative theory would not make evolutionary sense if arguments were addressed to people who were wholly unable to evaluate them from a sound epistemic perspective

R2.1. The double sided argumentative function of reasoning

M&S argue that communication has evolved to be advantageous to both communicators and receivers. Receivers receive rich info that they could not have obtained on their own, and communication allows communicators to achieve some desirable effect from receivers. M&S assert that they argue the main function of reasoning is social but is meant to serve the social needs of the individual in response to Dessalles’s concerns. M&S argue that receivers need to use epistemic vigilance to benefit from communication and agree with Opfer & Sloutsky that the main heuristic consists of assessing a communicator’s trustworthiness. M&S argue, however, that this is not the only heuristic used. Coherence checking is also important for receivers but can also be exploited by communicators. M&S concede that argumentation can be misused and abused to serve the interests of the communicator. This does not work, however, with receivers who care to be well informed. When people are motivated to reason, they do a better job accepting only sound arguments.

R2.2 Other functions of reasoning?

Many commentaries agree with argumentation but suggest that it may serve additional social functions or functions contributing to individual cognition. M&S recognize this possibility and explain their claim was that argumentation was the main function of reasoning but any evolved mechanism can be put to a variety of uses. Dessalles and Frankish suggest argumentation could have evolved as a means to display one’s intellectual skills. M&S agree argumentation could be put to such use but only occasionally, usually in academic milieus, and has actually evolved to be efficient rather than impressive. Pietraszewski distinguishes two classes of reasoning that show argumentation is not used just in the defense of factual claims but also of claims that are matters of choice or social alignment. M&S welcome this observation but argue it simply highlights that communication involves a mix of means and goals. Baumeister et al. draw attention to consciousness and culture. M&S acknowledge the need for more research on the connection between consciousness and reasoning but are not convinced culture contributes to the function of reasoning. Godfrey-Smith & Yegnashankaran suggest that reasoning is individualistic in function but dialogic in structure. Evans and Frankish also argue that reasoning has evolved to serve individual cognitive goals, including anticipating the future and strengthening resolve. M&S do not dispute these claims but dispute that individual cognition is the main function of reasoning. They argue that the main contribution of reasoning to individual cognition is in helping people evaluate other people’s arguments, and that argumentation is therefore the main function.

R3. Strength and biases of reasoning and argumentation

R3.1. Are we really good at argumentation?

In this section, M&S address commentaries that argue that argumentative skills can be improved with training and critique the data used by M&S as evidence of people’s basic argumentative abilities. Overall, M&S concede that spontaneous argumentation skills are imperfect and can be improved by teaching (which is linked to the variable importance given to argumentation in different cultures and institutions) but maintain that they display a “remarkable superiority” to the reasoning skill elicited in non-argumentative contexts.

R3.2. How efficient is group reasoning?

This questions elicited contrary opinions from commentators. M&S stress that the argumentative theory does not predict that groups will always make better decisions, but merely that reasoning should work better in context of a genuine debate. They agree that many aspects other than reasoning can impact the outcome of a discussion and that reasoning in a group can bring poor outcomes when there is no genuine deliberation. They also concede that sometimes the best arguments will point in the wrong direction.

R3.3. The strength of confirmation bias

M&S argue when we look for arguments in a debate, we are mostly interested in arguments for our side or against the other side, which is why they say confirmation bias is a feature of reasoning. Poletiek questions the evidence from hypothesis testing, which M&S understand, but emphasize that reasoning is still unable to correct our own intuitions even though it can easily try and correct those of others. Wolfe present studies on myside bias, which M&S argue merely reflects a belief that it is better to provide arguments for one’s side rather than also for the other side.  DeNeys and Stupple & Ball critique M&S’s interpretation of the belief bias data because some people engage in logical reasoning when faced with such problems. M&S agree that in reasoning tasks people try to provide the correct, logically valid answer, but it is interesting that most of them fail. They argue this indicates that reasoning is not geared towards pure logical validity.

R4. On the working reasoning

R4.1. The algorithmic level

M&S acknowledge that their theory has a limitation in that it does not address the implications on the algorithmic implementation of reasoning (Khlentzos & Stevenson). They appreciate the contributions made by commentators on this issue. Most notably, Weber & Johnson offer a process-level specification of how reasoning works in decision making. M&S argue that this theory predicts reason-based choice and confirmation bias. M&S believe this theory does not compete with argumentative theory because it is based on the workings, rather than the function, of reasoning.

R4.2. Reasoning outside the lab

M&S applaud Narvaez for pointing out the limitations of the target article’s focus on experiments carried out in the laboratory. Many argue that WEIRD people (Western educated industrialized rich democratic) behave differently than the rest of the world. But M&S argue that the available data do not show that a culture would be deprived of reasoning and argumentative skills. Even illiterate societies can solve logical problems in the proper contexts. M&S address the importance of developmental data in the study of argumentative theory but did not focus on it. Narvaez and Wu provide further support for the argumentative theory by drawing attention to the political sphere. M&S argue that their theory can explain both the successes and failures of political debates. M&S concede that their theory applies to scientists and philosophers, including themselves (Johnson). Finally, M&S argue that people should be somewhat receptive to moral arguments while evaluating them on the basis of their own moral intuitions.

R5. Conclusion

M&S explain that the commentaries have not led them to revise their theory in any major way, but have pointed to fascinating directions for future research. They concede that more needs to be done to link their ultimate theory with process theories of reasoning. M&S suggest that other mechanisms besides reasoning might benefit from being viewed as having a social function and hope their contribution contributes to the growing body of research that shows that the human mind is a social mind.  

Discussion Questions:

1. Are M&S not accommodating enough to other functions of reasoning, independent of whether or not the argument is the true main function?
2. Do you agree with Johnson that argumentative theory should also be applied to professional reasoners like scientists and philosophers? Are they “seeking truth” or merely building their own arguments?
3. De Neys asserts that just because people reason well in argumentative contexts doesn’t mean that they don’t try to reason logically outside of this context. How well, if at all, do you think people reason outside of an argumentative context?
4. Do you accept M&S’s dismissal of Narvaez’s concern with their focus on WEIRD people? How do you think testing non-WEIRD people’s argumentative skills would be different and how would the results impact M&S’s theory?

Introduction

Reasoning is a special kind of inference that consciously produces new mental representations from other consciously held representations, and it is unique to human beings. Mercier and Sperber specifically define reasoning as the production of a consciously produced conclusion from other consciously held premises (i.e. consciously held reasons), with an intuitive premise-conclusion component. In their article, they set out to investigate both how and why human beings engage in reasoning, although they focus most closely on the question of why (57).

 

1. Reasoning: Mechanism and Function

 

1.1 Intuitive Inference and argument

Mercier and Sperber set out to create their own dual-process approach to distinguish between intuitions (system 1 reasoning) and logical reasoning (system 2 reasoning): processes of inference and reasoning proper (58). The output of processes of inference are intuitive beliefs, that are formed at a “sub-personal,” unconscious level, for which were are not aware of the reasons behind holding them. The output of reasoning proper are reflective beliefs, that we do have conscious reasons for holding. Reasoning proper allows us to represent our own mental representations as well as the representations of others (metarepresentations). It does, however, contain intuitive elements on a fundamental level.

 

M&S give Decartes’ Cogito argument (“I think therefore I am”) as an example of reasoning proper (59). Although the thinker is able to give reasons for believing the argument that they exist, the fundamental reasons for accepting this as an intuitively good argument are much cloudier. In this case, the clear-cut dual system becomes less clear, because the outputs of what was originally considered pure system 2 reasoning must contain some elements of system 1 reasoning (intuition). The premise-conclusion relationships in an argument must be intuited on an unconscious level if one is to be able to evaluate the merit of an argument. The function of this ability is extremely evolutionarily salient.

 

1.2 The function of reasoning

M&S reject the classical view of the evolutionary function of reasoning as an enhancement of individual cognition. Under this classical view, system 2 reasoning is achieved by correcting mistakes in system 1 intuitions. They also reject the hypothesized function of reasoning as a mechanism by which organisms can react to novel environments favorably, which they argue is simple learning. Instead, they propose that reasoning evolved as a form of epistemic vigilance, which allows the senders and receivers involved in interpersonal communication to evaluate the information and arguments being exchanged (60).

 

This potential evolutionary origin of reasoning stems from the psychological concepts of trust calibration and coherence checking. Put simply, individuals must be able to quickly and accurately evaluate new information received from other individuals, to avoid being misled. With a mechanism that allows communicators to effectively communicate and evaluate new ideas, the information that humans are able to share increases in both quantity and epistemic quality. This ability would be strongly selected for in an environment in which the fast exchange of accurate knowledge is essential. In other words, the main function of reasoning is argumentative (60).

 

2. Argumentative Skills

 

2.1 Understanding and Evaluating Arguments

M&S state that there is a common conception that people in general are not very skilled arguers, a fact which would pose insurmountable obstacles to their theory that the primary function of human reasoning is argumentation (61). However, studies on persuasion and attitude change have shown that “when they are motivated, participants are able to use reasoning to evaluate arguments accurately.” Logical performance in reasoning research has been notoriously poor, but M&S argue this is due to the abstract, decontextualized nature of the tasks. In an argumentative context, people perform much better on the same kinds of tasks. Fallacies of argumentation are different than logical fallacies, and people generally perform well both in identifying argumentative fallacies and rejecting or accepting them as appropriate in context.

 

2.2 Producing Arguments

Previous studies would indicate people are generally unskilled in argument production as well. M&S argue that these apparent deficiencies actually stem from the context of the experimental tasks, which were ill-suited to reasoning’s argumentative function. In fact, people will use relevant data when they have access to it, they will develop more complex arguments if they anticipate any challenge to their assertions, and they are capable of formulating counterarguments so long as the argument being challenged is not their own (62).

 

2.3 Group Reasoning

M&S claim that prior reasoning research shows that in group settings, the dominant scheme is “truth wins.” Individuals show large improvements in reasoning tasks after group debates (an incredible increase of 10% to 80% correct responses in the Wason selection task) (63). Transcripts of group discussions, and the assembly bonus effect, suggest that this improvement is actually due to reasoning improvement, and not simply some members following other, “smarter” ones. M&S’s theory also predicts a strengthening of group opinion in artificial, nonoptimal group settings of prior agreement.

 

3. The Confirmation Bias: A Flaw of Reasoning or a Feature of Argument Production?

According to M&S’s theory, confirmation bias is not a flaw of reasoning but rather a feature. In some cases of “confirmation bias,” people are simply trusting their beliefs by drawing positive inferences from them; no proper reasoning has occurred. According to their theory, true confirmation bias should only occur in argumentative situations and only in argument production. It is not a general confirmation bias, but rather a bias toward confirming one’s own arguments and refuting those of others (64).

 

3.1 Hypothesis Testing: No Reasoning, No Reasoning Bias

M&S argue that the people’s poor performance in hypothesis testing is not due to reasoning at all. Lacking an argumentative setting, participants are simply adopting a “positive test strategy” in an intuitive manner. If, instead of producing their own hypothesis, people are given one from someone else, they do employ reasoning and are better able to falsify it.

 

3.2 The Wason Selection Task

M&S claim that poor performance on the Wason selection task occurs because the utterance of certain concepts in the in rule itself incites intuitive mechanisms of comprehension which cause people to focus on certain cards and make an inference as to the answer; subsequent reasoning processes only justify the answer already formulated. Participants in argumentative group settings perform much better on the task, as do people who are highly motivated to disprove the rule provided (65).

 

3.3 Categorical Syllogisms

People perform poorly on categorical syllogisms because solving them correctly requires producing counterexamples to one’s own conclusion. They perform much better if the conclusion is unbelievable or if it is provided by someone else.

 

3.4 Rehabilitating the Confirmation Bias

The confirmation bias is traditionally viewed as a dangerous defect in reasoning, due to cognitive limitations. The fact that people are quite adept at falsifying propositions when motivated troubles the matter of causation, and the consequences of confirmation bias are only disastrous in abnormal contexts of prior agreement (whether inter- or intra-personally). In more felicitous contexts of groups solving disagreements, the confirmation bias is an efficient division of cognitive labor (65). High performance in group reasoning tasks suggests that the confirmation bias is primarily present in argument production, not evaluation.

 

4. Proactive Reasoning Belief Formation

Most of our beliefs as individuals go unchallenged as they are unexpressed or only relevant to ourselves. If we identify a particular belief of ours as possibly contentious, we regard it as an opinion and may search proactively for arguments to justify this opinion, a phenomenon studied as “motivated reasoning.” (66)

 

4.1 Motivated Reasoning

In a study in which participants were given a fake medical result, they tended either to discount the rate of false positives provided, or utilize it to undermine the test, depending on whether their result was positive or negative. If this were due to wishful thinking, participants could dismiss the test entirely, but instead they produced arguments to support their opinion. To M&S, this motivated reasoning is targeted at justifying beliefs to others; any personal belief revision in the name of truth-seeking is incidental.

 

4.2 Consequences of Motivated Reasoning

 

4.2.1 Biased Evaluation and Attitude Polarization

When participants are presented with a study either confirming or attacking their prior position on the death penalty, they are more likely to criticize the methodology if the conclusion reached differs from their own. M&S interpret this as evidence that people’s goals in such a situation are “argumentative rather than epistemic.” (67) Additionally, people spend more time evaluating an argument contrary to their own opinion, as rejecting the argument requires justification that accepting it does not.

 

4.2.2 Polarization, Bolstering, and Overconfidence.

When people think about a stimulus regarding which they have a prior decided opinion, they tend to polarize and strengthen their existing attitude, rather than reevaluating it. This tendency increases with time spent thinking, motivation to think, and the number (if any) of explicit arguments the person puts forth supporting their opinion (67). Being publicly committed to the opinion results in bolstering, an increased pressure to justify the opinion rather than change it, an effect which is strengthened by heightened accountability. Providing an answer to a question causes people spontaneously to produce justifications for their answer, resulting in subsequent overconfidence.

 

4.1.3 Belief Perseverance

Belief perseverance depends on the orientation of people’s intuitive inferences, and whether evidence presented supports these inferences, rather than on the order of evidence presented, indicating that belief perseverance is simply a special type of motivated reasoning (68).

 

4.1.4 Violation of Moral Norms

The study of moral hypocrisy shows that reasoning is better suited to justifying people’s actions than guiding them (serving an argumentative rather than moral or epistemic goal) (69). The effect of moral hypocrisy in certain judgments can be eliminated by introducing cognitive load during the judgment process and therefore interfering with reasoning.

 

5. Proactive reasoning in decision making

Mercier and Sperber argue that the main role of reasoning is done in anticipation in anticipation of the necessity of defending a decision. This process does not always result in the weighing of pros and cons in an reliable way (the classical view of reasoning), as has been shown by extensive empirical evidence (69).

 

5.1 To what extent does reasoning help in deciding?

Many studies have shown that decisions based on careful, conscious reasoning actually results in poorer decisions than does unconscious decision-making processes that are not based on carefully stated reasons (69). Most decisions are made intuitively, and those that are made through reasoning often result in decisions that may be easy to justify, but they may not be the best decisions (69).

 

5.2 Reason-based choice

This bias that people have to make more readily justifiable decisions causes them to make a number of classically “irrational” decisions, to avoid the risk of criticism. This phenomenon, termed reason-based choice, causes people to make “mistakes” on tasks designed to measure rationality. Making choices based on reasons, no matter the justifiability of those reasons, will be favored when making decisions and thus result in irrational decisions (70).

 

5.3 What reason-based choice can explain

Reason-based choice, M&S argue, can explain a great number of the well-known challenges to human rationality. They list the disjunction effect, the sunk-cost fallacy, framing, and preference inversion as empirical psychology research examples of reason-based choice in action (70). What all of these examples have in common is that they provide significant evidence for cognitively unsound uses of reasoning (71). M&S define these deviations from irrationality as the misuse of an evolutionarily favorable mechanism for decision making. As they argue throughout the text, reasoning most likely evolved to function in a social context, and allows for people to anticipate which arguments they need to justify in order to have other people take their beliefs seriously. At its core, their argument is that the function of reasoning is to lead people to justifiable decisions, and not necessarily good decisions (as defined by a classical definition of rationality). The instances in which this distinction must be made (i.e. between justifiable and good) are rare, and therefore do not pose a significant threat to their argument.

 

6. Conclusion: Reasoning and rationality

Reasoning allows for human communication to be both reliable and potent, and benefits both senders and receivers in the exchange of information. This argumentative theory of reasoning shows that “irrationality,” as it is classically understood in psychology and philosophy, is merely the result of the human tendency to systematically look for arguments to justify beliefs and actions. As Mercier and Sperber demonstrate with their review of the results of many reasoning tasks, people engaged in argumentation favor arguments that support their own views if they have “an axe to grind” about the argument, but truth wins when all participants have equal interest in discovering the right answer to a problem (72). This means that truth doesn’t necessarily always win, but the best arguments do. With time and with enough participants engaged in conversation, however, the best arguments will eventually equate to the truth.

 

Discussion questions:

1. Does an argumentative function of reason have disastrous moral or epistemic consequences?
2. Can the biased features of argumentative reasoning be effectively modulated by group debate or is another solution in order?
3. How does Mercier and Sperber’s model of reasoning differ from complex learning mechanisms? How might this be selected for in an evolutionary context?
4. Does this conception of reasoning fit into a normative or a descriptive model of rationality?

Open Peer Commentary:

Throwing the normative baby out with the prescriptivist bath water

Achourioti, Fugard, &  Stenning agree with E & E that it would be foolish to say there is one formal model of rationality, but they disagree with the argument that descriptivism is the answer. These authors argue that there are two norms that are integral to the process of human reasoning:

Constitutive norms: These norms answer the question of “what the reasoning is”

Regulative norms: These norms answer the question of “why the reasoning is the one that it is”

These authors argue that “thoroughgoing descriptivism” is not enough to explain how people reason and the level of people’s understanding, and normative concepts cannot be extricated from discussions about human reasoning.

 

Norms for reasoning about decisions

Bonnefon addresses the new “trend” in philosophy to look at how reasoning intersects with decision-making. Bonnefon shows three different examples of how this new intersection provides an opportunity for both normativist and descriptivist approaches. Bonnefon argues that although these two branches of philosophy (decision-making and reasoning) rely on mostly normativism as independent subjects, when these subjects are combined a strict normativist approach is weaker.

 

The unbearable lightness of “Thinking”: Moving beyond simple concepts of thinking, rationality, and hypothesis testing

Brase & Shanteau argue that the HTT is not sufficient to solve the problems posited by E & E and instead the focus should be on “a better conception of how theories are constructed and evaluated” (250). The method of “strong inference” in which tests are used to compare many “viable” hypotheses to see which can be excluded should be preferred. There should not be one model that encompasses how thinking works. Rather, Brase & Shanteau argue that there should be an emphasis on the “domain-specificity” of reasoning, or the concept that there are many different types of “thinking.”

 

Competence, reflective equilibrium, and dual-system theories

Buckwalter & Stich agree with E & E on the merits of descriptivism, but these authors argue that the distinction between competence theory and normative theory is not well defended. John Rawls posits the idea of a “reflective equilibrium” in which moral principles and moral judgments align. Cohen furthers this concept with respect to reasoning and says that if there are normative and descriptive models of reasoning, these models must “coincide” (252). Buckwalter & Stich argue that if the dual process model of reasoning is correct, it is unlikely that a normative model could ever be correct. System 2 seems to be a system that could vary individual to individual and therefore cannot be encompassed by one model.

 

A role for normativism

Douven argues that there are merits to both normativism and descriptivism in conversations about human reasoning. Douven argues that “long-run accuracy” is a priori our “epistemic goal,” and that E&E completely miss this point in their argument (252). Douven also disagrees with E & E’s argument that empirical research is not helpful in order to distinguish between models of reasoning. Douven argues that empirical research is helpful in the case when there is no known model for a type of reasoning. Douven believes that E&E’s argument is not strong enough to entirely throw out normativism.

 

The historical and philosophical origins of normativism

Novaes is interested in the question “is thinking a normative affair at all?” (253). Novaes points to the history of the thought that thinking falls into the category of normativism and that logic should be the normative system used. Novaes draws the connection to Kant’s philosophy, a perspective that is defined by “transcendental idealism” (254). Novaes states that if we reject transcendental idealism, then we can also reject the concept that thinking is normativist.

 

Just the facts, and only the facts, about human rationality?

Foss is in agreement with E&E that there should be greater emphasis on scientific facts. However, Foss argues that they failed to give a definition of rationality in their argument. In addition, Foss argues that competency theory is a type of normativism, and so it may be that normativism cannot be entirely rejected in psychological studies.

 

Overselling the case against normativism

Fuller & Samuels agree with E&E’s argument but find flaws in two aspects of their argument. Firstly, Fuller & Samuels think that researchers will end up falling into “normative interpretations” even if they rely on formal theories (255). Secondly, Fuller & Samuels argue that E&E have a too narrow definition of normativism.

 

Undisputed norms and normal errors in human thinking

E&E state that there seem to be multiple norms for any given task of human reasoning and therefore it cannot be determined which norm is correct. Girotto disagrees with this argument and states that there often can be one norm, and that norms are useful in providing guidance for individuals. Therefore, norms are necessary in conversations about human thinking.

 

Normative theory in decision making and human reasoning

Gold, Colman, & Pulford argue that the is-ought problem is not significant enough to reject normativism. Rather normative theories can be used in “generating powerful descriptive theories” (257). In conversations about moral reasoning and morals in general, norms are incredibly useful. Therefore, normativism should not be completely thrown out.

 

Why rational norms are indispensable

Hahn argues that norms are important because they provide standards that can be used as “interpretative tools”, with respect to evaluation and prediction of behavior (257). Hahn argues that these normative models would be well complemented by descriptivism. Hahn also argues that the “is-ought” distinction has nothing to do with normativity whatsoever. Therefore, normativism is a crucial model and should not be rejected.

 

Defending normativism

Hrotic, an anthropologist, shares his concern that “in practice, the distinction between ‘oughts’ is fuzzy” (258). He raises an important question on how necessary, if at all, it is to be fully aware of when you are using an directive ought vs. an evaluative ought. He goes on to discuss how biases are relevant to and, perhaps, also useful in understanding academic methods/human reasoning despite our lack of ability to articulate the reasons for our biases.

 

Cultural and individual differences in the generalization of theories regarding human thinking

Kim & Park agree with E&E’s argument in favor of descriptivism over normativism, but argue that the current form of descriptivism is still limited in allowing us to fully comprehend human cognitive processes. Kim & Park offer us a way to address this limitation. They emphasize the importance of including cultural differences when conducting descriptivism research, suggesting that individual differences across different cultures influence behaviors unique to a particular group or culture. Kim & Park also suggest that while cognitive goals may differ across cultures, the motivational system (goal activation) by which the goals are pursued and achieved may be universal (pg 259-260).

 

Norms and high-level cognition: Consequences, trends, and antidotes

Though McNair & Feeney agree with E&E, they are less critical in their objections towards normativism. The authors have three main points. 1) Normativism is not uniformly disastrous as normativism has enabled us to understand behaviors that are based in normative models (Bayesian). 2) Normativism in the debate on human reasoning will continue to be increasingly relevant. 3) Primarily focusing on expert reasoners in studying human cognition is limiting and problematic. McNair & Feeney note that it is imperative for us to find a way to study human reasoning processes across both naive and expert reasoners.

 

Norms, goals, and the study of thinking

Nickerson argues that there is a need to find balance between normativism and descriptivism. Though he gives merit to E&E’s efforts to challenge normativism and their suggestion that research should focus on “how thinking is actually done” (261), Nickerson finds is less persuaded by E&E’s attempt to dismiss normativism completely from the field. Nickerson proposes that a middle ground must be established between the two different approaches. He suggests that descriptivism is used to learn how reasoning is done in order to understand how we ought to reason.

 

The “is-ought fallacy” fallacy

Oaksford and Chater claim that E&E’s “application of the “is-ought” fallacy is itself fallacious” (262). O&C clarify that their original argument in their paper did not claim that the Bayesian probability ought be the normative system by which we follow – a claim they say E&E assume O&C made. O&C also discuss how certain statements of explanation do not directly lead to an “ought” statement, but rather derive what “is” – stating descriptive facts by means of normative theories.

 

Systematic rationality norms provide research roadmaps and clarity

Pfeifer argues that normative theories should not be eliminated from cognitive research. Pfeifer claims that E&E’s argument that “conditional elimination inferences are single-norm paradigms” (263) fails to be free of conflict in the context of probability logic. The author suggests that instead of doing without normativism entirely in psychological research, as E&E proposes, possible improvements/changes to current normative theories should be considered.

 

A case for limited prescriptive normativism

Pothos and Busemeyer argue that identifying when a certain cognitive process was successfully applied is strictly contextual. They go on to explore Quantum probability and how some researchers have claimed that this model is a foundation for understanding human reasoning/thinking. P&B conclude that it is quite impossible to conduct valuable research on cognitive processes in the absence of formal frameworks and that such use of formal frameworks “inevitably… lead to some limited prescriptive normativism” (265).

 

Epistemic normativity from the reasoner’s viewpoint

Proust argues that E&E did not attempt to fully consider how an individual’s ability to assess his or her own reasoning performance impacts how he or she carries out first-order reasoning tasks. Proust suggests that there is variability in the way one person approaches a particular task than another person does, establishing that reasoners each interact with tasks differently due to his/her own respective set of experiences.

 

Naturalizing the normative and the bridges between “is” and “ought”

Quintelier and Fessier draw attention to the negative consequences that can possibly result from applying E&E’s approach to fields of scientific research beyond the cognitive sciences. They note that the meaning of a normative term very much depends on “one’s epistemological or meta-ethical views” (266). Q&F offer us two naturalistic ethics examples that walk us through ways in which “ought” is used in proper context to address a particular and relevant conclusion to a particular task.

 

Truth-conduciveness as the primary epistemic justification of normative systems of reasoning

Schurz appreciates E&E’s efforts to challenge normativism, but identify that E&E’s argument falls short. Schurz bases his response to E&E on two main points: 1) that formality is not synonymous to normativism and 2) that evaluative norms are in no need of justification. Schurz emphasizes the importance of recognizing that a form of prescriptive normativism is necessary in our understanding of the human rational and psychology.

 

Reason is normative, and should be studied accordingly

Spurrett rejects E&E’s argument and claims that it is “always appropriate” to assess whether an individual’s reasoning reflects “truth” (268). He demonstrates how people can and do reason poorly, ultimately concluding that it is “nonsensical” to define these processes as reasoning in the way that a purely descriptive approach would require, and criticizes the authors of the target article for their vague and seemingly contradictory definition of normativism.

Normative models in psychology are here to stay

At the heart of Stanovich’s argument is the idea that normative theory has been, and continues to be, a useful metric in the study of human reasoning (268). He rips into what he sees as an artificial division between Bayesian theories and instrumental rationality, which is essential to E&E’s model but contradicts the prevailing idea that the latter incorporates the former. Stanovich also emphasizes the fact that people tend to be able to recognize that a normative strategy is a superior solution to a problem (269).

Understanding reasoning: Let’s describe what we really think about

Sternberg appreciates the underlying motivation behind the target article and expands on it, arguing that normative models have caused psychologists to focus on a “narrow sliver” of the problems and decisions that people commonly face (270).

Normative benchmarks are useful for studying individual differences in reasoning

Stupple and Ball are sympathetic to E&E’s point that matching reasoning decisions to seemingly consistent normative strategies can lead to an incorrect diagnosis of the “underlying analytic process” (270). However, they ultimately argue that such “normative responses” can be incredibly useful to researchers, as demonstrated by their proposed methodological triangulation approach (271), and that normativism and descriptivism are on a continuum with no clear division.

Probability theory and perception of randomness: Bridging “ought” and “is”

Sun and Wang are skeptical of E&E’s call to abandon normativism, which they claim to be as valuable as descriptivism to the study of streak patterns in Bernoulli trials (271), but agree that researchers should not throw around “ought inferences.”

Normativism versus mechanism

Thompson defends E&E’s position on the role of normative models in psychology, emphasizing that normativism is misleading and limited (272). Her main point is that it causes researchers to focus on whether reasoning is “good” or “bad” rather than explore the complicated processes that dictate reasoning.

Neurath’s ship: The constitutive relation between normative and descriptive theories of rationality

Waldmann acknowledges that it would problematic to overstate how well a normative model fits empirical results, but thinks that normativism is very relevant to theoretical, causal, and practical rationality when used correctly (273).

What is evaluative normativity, that we (maybe) should avoid it?

Weinberg takes issue with E&E’s assertion that psychological theory should not “substantively” incorporate evaluative normativity, arguing that normativism is already too well incorporated into current psychology (274).

Authors’ Response: Towards a descriptivist psychology of reasoning and decision making

R1. Introduction

E&E reiterate two of their underlying arguments: that “is-ought” inferences based on normative theories are not reliable and that approaching psychological theory from a normativist viewpoint leads to “systematic biases” in current research (275).

 

R2: Between normativism and descriptivism: Definitions and boundaries

Instrumental rationality is not normative, and there is no clear line between the “ought”s of evaluative and instrumental. If formal theories are seen as simply computational, then we only give up evaluation when dismissing normativism. An alternative to descriptivism could be “soft normativism”, with normative evaluations in addition to goals of descriptive research. We need to distinguish normative and competence theory but still value why they are linked. Formal theories are levels relating to computational analysis (descriptive). Only used HTT as an example of descriptive theory that uses formal theories, so assumption of ought to adapting behavior to environment can be seen as normative. Bounded rationality cannot really fit with radical types of normativism.

R3: Epistemic rationality and self-knowledge

Psychological science should not focus on norms—they should spend time on conducting research. Constitutive rules are not normative, as they define the system (think chess rules) and what it is, not what it ought. Regulative rules regulate behavior (think table manners). The authors believe that we do not need a normative account of belief (think ranges in memory and vision). We can have descriptivist view on people’s own interpretations of their epistemic goals and norms, but what they deem as correct does not necessarily need to be normative, only plausible. Fast intuition leads to confidence in righteous behavior, which can be separate from it being normatively accurate. Justification does not equate to rationale: no one acknowledges previous biases in the reasoning for their actions.

R4: Normativism and descriptivism in dual-process research

Two minds theory: old (intuition, reinforcement of behavior) and new (reflective, goal oriented) mind both have control over our brain, sometimes cooperatively, sometimes competing. Abstract/normative does not equal Type 2 thinking, contextual/cognitive biases do not equal Type 1. There cannot be a normative reasoning with two separate systems, especially when the second one varies based on a person’s environment and identity.

R5: The new paradigm psychology of reasoning

The deduction paradigm has been changing through a Kuhnian revolution and paradigm shift as there is more of a push towards the integration of reasoning and decision making theories. The “new” paradigm psychology of reasoning removes truth and deduction as constraints, focusing more on probabilistic and Bayesian approaches, pragmatic factors, and degrees of belief. However, the new paradigm is divided on support for versus against the use of a normative framework, so the most popular alternative is Bayesianism.

        Bayesianism allows for inferences on beliefs with varying degrees of certainty as has a direct connection between reason and decision. The question then is whether it is a prescriptive normative account or descriptive. The authors see Bayesianism as an accurate descriptive account. Beliefs and utilities are more apt to be subjective based on individual accounts, making descriptive Bayesianism hard to disprove. Researchers have leaned toward alternative descriptive accounts because of changes in belief in the beginning compared to the end, unlike the Bayesian view of belief not changing simply based on sequence (example of evidence presented in a courtroom).

R6: Cognitive variability

Individual differences tend to be associated with normativist, and cultural differences with moderate relativism. There are individual and cultural differences in reasoning, e.g. age, IQ, working memory, Western vs eastern culture. Language is cognitively variable, making cultural norms of rationality based on language more apt. Cross cultural research is important in order to rid of previous researcher cultural biases.  Assessment of normative behavior shows change in what people do, whereas interpretation of the correctness of that behavior varies. Descriptivist approach tends to show qualitative differences. “Can” vs “cannot” has unclear boundaries, as ought and can cannot imply each other as well as Kant may have originally thought.

R7: Descriptivism versus normativism in conduct of empirical research on thinking

Normativism and descriptivism can play roles in research, but despite biases, normativism can also be valuable. There is support for looking at how we reason AND how we should. There is a difference between what we expect to be normative and what the behavior actually is. The authors are not sure if some of the arguments are based on normative accounts or competence/computation.

R8: Conclusions

The authors suggest any researcher take a step back before evaluating what they are doing. Regardless of whether there is progression towards a more descriptivist approach, the authors hope that they have at least provided a space for thought about reasoning and decision making, which have had normative theory in them, and conversation acknowledging normativism along the way.

 

Discussion questions:

1. Do you agree with Hahn that normativism has nothing to do with the is-ought distinction?
2. Sternberg argues that normative models are not applicable to the vast majority of everyday decisions. Is this an accurate assessment, and does it change how you feel about E&E’s argument?
3. How are Oaksford & Chater proposing the process of iteration? Do you think this is rational when looking at behavior and what is versus what ought?
4. Is it possible to find a middle-ground/balance between normativism and descriptivism when trying to understand how it is we think and how we ought to think? Nickerson suggests that the “truth is somewhere in between” the two methods. Do you agree?