Abstract

Inferring knowledge is a core aspect of human cognition. We can form complex sentences connecting different pieces of information, such as in conditional statements like “if someone drinks alcohol, then they must be older than 18.” These are relevant for causal reasoning about our environment and allow us to think about hypothetical scenarios. Another central aspect to forming complex statements is to quantify about sets, such as in “some apples are green.” Reasoning in terms of the ability to form these statements is not yet fully understood, despite being an active field of interdisciplinary research. On a theoretical level, several conceptual frameworks have been proposed, predicting diverging brain activation patterns during the reasoning process. We present a meta-analysis comprising the results of 32 neuroimaging experiments about reasoning, which we subdivided by their structure, content, and requirement for world knowledge. In conditional tasks, we identified activation in the left middle and rostrolateral pFC and parietal regions, whereas syllogistic tasks elicit activation in Broca's complex, including the BG. Concerning the content differentiation, abstract tasks exhibit activation in the left inferior and rostrolateral pFC and inferior parietal regions, whereas content tasks are in the left superior pFC and parieto-occipital regions. The findings clarify the neurocognitive mechanisms of reasoning and exhibit clear distinctions between the task's type and content. Overall, we found that the activation differences clarify inconsistent results from accumulated data and serve as useful scaffolding differentiations for theory-driven interpretations of the neuroscientific correlates of human reasoning.

INTRODUCTION

A core ability of humans is to represent and infer knowledge. The cognitive process of reasoning allows to transform implicit information into explicit one and to clarify what necessarily follows from given information. For example, take the following conditional statement: “If I eat too much sugar, then I will gain weight” and the observation: “Actually, I do eat too much sugar.” A human reasoner can easily infer that “I will gain weight.” This is because the second statement includes information about the actual condition at hand. The reasoning process lies in combining these different pieces of information: the conditional statement and the actual observation. Many scientific observations and everyday commonsense knowledge can be formulated by an “if–then” statement. Conditional tasks involve a conditional rule, that is, a rule that only holds if a precondition is met. A well-known example for these tasks is the Wason selection task (Wason, 1966). The task is applied with different content types, but the basic structure remains: A conditional statement “If P, then Q” is given, and the participant has to select the options that are sufficient to discount the rule from the following options: “P, Not-P, Q, Not-P.” Depending on the factual statement, four inference patterns are possible for a given conditional (see Table 1): two logically valid inference patterns called Modus Ponens and Modus Tollens (MT) and two logically invalid ones, Affirming the Consequent and Denying the Antecedent. Table 1 contains an example of a quantified reasoning task, a syllogism: “All bakers are chess players. All chess players are musicians. Hence, all bakers are musicians.” In syllogistic reasoning, one has to reason about the membership of objects to different categories or sets and infer group entailments (the set of all bakers, the set of all chess players, the set of all musicians). Based on the four quantifiers, “all” (A), “none” (E), “some” (I), and “some not” (O), there are typically 64 distinct syllogisms that can be deduced from the combination of the two premises.

Table 1. 
Examples of Conditional and Syllogistic Tasks
TaskExampleAbstraction
Modus Ponens If I eat too much sugar, I will gain weight. p → q 
I eat too much sugar. ∴p 
I will gain weight. 
  
Modus Tollens If I eat too much sugar, I will gain weight. p → q 
I did not gain weight. ¬q 
I did not eat too much sugar. ∴¬q 
  
Affirming the Consequent If I eat too much sugar, I will gain weight. p → q 
I have gained weight. ∴q 
I ate too much sugar. 
  
Denying the Antecedent If I eat too much sugar, I will gain weight. p → q 
I did not eat too much sugar. ∴¬p 
I did not gain weight. ¬q 
  
AAA Syllogism All bakers are chess players. ∀x Bx → Cx 
All chess players are musicians. ∀Cx → Mx 
All bakers are musicians. ∀B Bx → Mx 
TaskExampleAbstraction
Modus Ponens If I eat too much sugar, I will gain weight. p → q 
I eat too much sugar. ∴p 
I will gain weight. 
  
Modus Tollens If I eat too much sugar, I will gain weight. p → q 
I did not gain weight. ¬q 
I did not eat too much sugar. ∴¬q 
  
Affirming the Consequent If I eat too much sugar, I will gain weight. p → q 
I have gained weight. ∴q 
I ate too much sugar. 
  
Denying the Antecedent If I eat too much sugar, I will gain weight. p → q 
I did not eat too much sugar. ∴¬p 
I did not gain weight. ¬q 
  
AAA Syllogism All bakers are chess players. ∀x Bx → Cx 
All chess players are musicians. ∀Cx → Mx 
All bakers are musicians. ∀B Bx → Mx 

Furthermore, reasoning problems have been differentiated based on the content present in the premises. The differentiation in content types task has been widely discussed in the reasoning literature (Goel, 2007, Goel, Makale, & Grafman, 2004; Goel, Buchel, Frith, & Dolan, 2000). Most commonly, divisions are made between tasks featuring abstract (e.g., letters, nonintelligible words) and content-based tasks (tasks with contents having relevance in the real world). This differentiation is especially used when testing dual process accounts of reasoning, as the underlying assumption is that the differences in the tasks' content triggers diverging reasoning processes. For abstract tasks, this would be focusing on the logical structure of the task, whereas for content-laden tasks, a heuristic content-entailing mechanisms is assumed to be triggered (Goel et al., 2000). The successful solving of content-based tasks may require the application of world knowledge. World knowledge denotes the knowledge we assume to have about the state of the world. This term stems from a linguistic differentiation in which the branch of pragmatics, as opposed to semantics (the knowledge about the meaning of the words per se), is specifically concerned with knowledge about the internalized representation of the real world (Hagoort, Hald, Bastiaansen, & Petersson, 2004). In contrast, the content of abstract tasks does not require the recollection of knowledge about the world, but only of the semantic knowledge of the words. Hence, the tasks are asked to be evaluated and solved solely based on their immediate structure.

Current Psychological and Neurocognitive Findings on Human Reasoning

There are currently four main classes of cognitive theories about reasoning that are relevant for neuroscientific studies: theories assuming that reasoning is accomplished by building a mental representation, theories hypothesizing that language- and/or rule-based processes are employed during reasoning, probabilistic accounts proposing that reasoning processes are accomplished by probabilistic inference (Heit, 2015; Oaksford & Chater, 2001, 2007), as well as dual process accounts assuming two processes enabling reasoning (Goel, 2003). Mental model theory (Johnson-Laird, 1980, 1983, 2006) is a model-based account. Assuming that, after the comprehension of the given information, an analogical mental representation of the situation—a mental model—is constructed. This model is inspected and, if necessary, fleshed out to search for possible counterexamples. Subsequently, the maintenance and manipulation of a mental representation require corresponding neural activation in regions like the parietal cortex (Knauff, Fangmeier, Ruff, & Johnson-Laird, 2003). In contrast, language- and rule-based approaches, such as mental logic theory, assume syntax-based information processing, including the application of logical rules that reasoners have at their disposal (e.g., as formulated in the mental logic theory; Rips, 1994, 2001; Braine & O'Brien, 1991; Braine, 1978; Henle, 1962). From a neurocognitive perspective, this view necessitates the processing of syntactical information and may require the activation of syntax-related brain regions (Knauff et al., 2003). Following from these theoretical conceptualizations, a “new paradigm” of human reasoning emerged in the last two decades, including dual process theory as well as probabilistic accounts. Dual-process models of reasoning propose that reasoners apply different task-solving strategies that can be largely divided into System I processing characterized as automatic and associative and System II, which is deliberate and analytic (e.g., see Stanovich & West, 2000; Evans & Over, 1996). As previously mentioned, the processes are supposed to be triggered by the presentation of the task and the conveyed content. Probabilistic accounts of reasoning are based on the assumption that human reasoning processes are based on probabilities; hence, reasoning can be characterized as a Bayesian process (Oaksford, 2015; Oaksford & Chater, 2001, 2007). Although this paradigm has been used to reinterpret and criticize meta-analyses of neuroimaging studies on reasoning, it does not make distinct predictions about brain activation during the reasoning process (Oaksford, 2015).

Extensive analyses of behavioral data and cognitive theories for syllogistic reasoning (e.g., Khemlani & Johnson-Laird, 2012) and conditional reasoning (e.g., Oberauer, 2006) do support a slightly higher predictive ability of model-based approaches in contrast to alternative non-model-based theories for both domains. The main question is if these fundamental differences are reflected in the neuronal processes of reasoning, as well with consistent activation in regions associated either conceptual framework. Such findings can support either theory on the level of neural implementation. Consequently, a first comprehensive analysis of neuroimaging studies has been conducted about a decade ago by Prado, Chadha, and Booth (2011) in which the neural correlates of deductive reasoning were investigated in general, as well as relational, propositional (including conditional), and categorial (including syllogistic) reasoning in specific. Across all experiments, they identified a bilateral but rather left-centered frontoparietal network for general deductive reasoning, as well as the left precentral gyrus (PreCG) and posterior parietal cortex (PPC) during conditional reasoning and the left inferior frontal gyrus (IFG), PreCG, and bilateral BG during syllogistic reasoning.

The Need for a New Meta-analysis

The study by Prado et al. (2011) provided important insights into general mechanisms of human reasoning and allowed to distinguish between the three domains of reasoning. However, it only consisted of 12 experiments for conditional and eight experiments for syllogistic reasoning. Since 2011, the number of experiments that we can incorporate in our study significantly increased: We were able to identify 16 and 17 experiments, respectively. Moreover, by task analysis, we concluded that conditional and syllogistic reasoning are both broad categories subsuming dissimilar problem formulations under the respective umbrella term. Hence, there is a need for refinement. This was already proposed by Prado et al. (2011) by arguing for a more fine-grained analysis, especially with respect to conditional reasoning, as the aggregated conditional studies did not exhibit a comprehensible picture of the underlying cognitive processes, which they attributed to the heterogeneity of the task designs involved.

METHODS

Our aim is to thoroughly investigate the neural correlates of human reasoning. We conducted an extended literature research of neuroimaging studies examining the cognitive mechanisms subserving the solution of cognitive reasoning tasks using conditional and syllogistic statements. We aimed at including as many experiments as possible to ensure statistically robust and reliable results.

Article Acquisition and Selection

For data acquisition, we considered the aggregate study by Prado et al. (2011) to find relevant neuroscientific articles. Furthermore, we conducted online searches via the platforms PubMed, ScienceDirect, and Google Scholar for acquiring studies published until 2019. Our inclusion criteria were the following: conditional or syllogistic reasoning problems solved by healthy participants. All data points yielded by fMRI or PET needed to be reported in either MNI (Collins, Neelin, Peters, & Evans, 1994; Evans et al., 1993) or Talairach space (Talairach & Tournoux, 1988) and acquired by whole-brain analyses. The studies needed to be published in peer-reviewed scientific journals.

Article Categorization

After having acquired and preselected the articles (for a comprehensive overview, see Table 2), we applied a task categorization scheme to efficiently analyze the data and draw more detailed conclusions from subsequent results. Concerning the results, we conducted the analyses for all subsequently mentioned subsets, as well as conjunction and contrast analyses for the inference type (syllogistic/conditional), the content, and the world knowledge division. In the following, we will elaborate on the applied contrasts.

Table 2. 
Overview of the Experiments Included in the Meta-analysis
Reasoning TypePublicationYearSubjectsWorld KnowledgeContent
Conditional Parsons & Osherson 2001  10 NWK Yes 
Noveck, Goel, & Smith 2004  16 NWK No 
Canessa et al. 2005  12 NWK Yes 
Monti, Osherson, Martinez, & Parsons, Exp 1 2007  22 NWK No 
Monti et al., Exp 2 2007  22 NWK Yes 
Prado & Noveck 2007  20 NWK Yes 
Reverberi et al. 2007  14 NWK No 
Monti, Parsons, & Osherson 2009  15 NWK No 
Prado, Van Der Henst, & Noveck 2010  13 NWK No 
Reverberi et al. 2010  26 NWK No 
Liu et al. 2012  14 NWK No 
Canessa, Pantaleo, Crespi, Gorini, & Cappa 2014  14 NWK Yes 
Hearne, Cocchi, Zalesky, & Mattingley 2015  21 NWK No 
Baggio et al. 2016  30 NWK No 
Coetzee & Monti 2018  20 NWK No 
  
Syllogistic Goel, Gold, Kapur, & Houle 1997  10 WKA Yes 
Goel, Gold, Kapur, & Houle 1998  12 NWK No 
Osherson et al. 1998  10 NWK Yes 
Goel et al. 2000  11 WKA, NWK Yes, No 
Goel & Dolan 2003  14 WKA, NWK Yes, No 
Goel & Dolan 2003  19 WKA Yes 
Goel & Dolan 2004  16 WKA Yes 
Kroger, Nystrom, Cohen, & Johnson-Laird 2008  16 NWK Yes 
Rodriguez-Moreno & Hirsch 2009  12 NWK Yes 
Jia et al. 2009  11 WKA Yes 
Reverberi et al. 2010  26 NWK No 
Prado et al. 2010  13 NWK No 
Reverberi et al. 2012  26 NWK No 
Smith, Vartanian, & Goel 2014  16 NWK Yes 
Brunetti et al. 2014  13 NWK Yes 
Porcaro et al. 2014  13 WKA Yes 
Smith, Balkwill, Vartanian, & Goel 2015  17 WKA Yes 
Reasoning TypePublicationYearSubjectsWorld KnowledgeContent
Conditional Parsons & Osherson 2001  10 NWK Yes 
Noveck, Goel, & Smith 2004  16 NWK No 
Canessa et al. 2005  12 NWK Yes 
Monti, Osherson, Martinez, & Parsons, Exp 1 2007  22 NWK No 
Monti et al., Exp 2 2007  22 NWK Yes 
Prado & Noveck 2007  20 NWK Yes 
Reverberi et al. 2007  14 NWK No 
Monti, Parsons, & Osherson 2009  15 NWK No 
Prado, Van Der Henst, & Noveck 2010  13 NWK No 
Reverberi et al. 2010  26 NWK No 
Liu et al. 2012  14 NWK No 
Canessa, Pantaleo, Crespi, Gorini, & Cappa 2014  14 NWK Yes 
Hearne, Cocchi, Zalesky, & Mattingley 2015  21 NWK No 
Baggio et al. 2016  30 NWK No 
Coetzee & Monti 2018  20 NWK No 
  
Syllogistic Goel, Gold, Kapur, & Houle 1997  10 WKA Yes 
Goel, Gold, Kapur, & Houle 1998  12 NWK No 
Osherson et al. 1998  10 NWK Yes 
Goel et al. 2000  11 WKA, NWK Yes, No 
Goel & Dolan 2003  14 WKA, NWK Yes, No 
Goel & Dolan 2003  19 WKA Yes 
Goel & Dolan 2004  16 WKA Yes 
Kroger, Nystrom, Cohen, & Johnson-Laird 2008  16 NWK Yes 
Rodriguez-Moreno & Hirsch 2009  12 NWK Yes 
Jia et al. 2009  11 WKA Yes 
Reverberi et al. 2010  26 NWK No 
Prado et al. 2010  13 NWK No 
Reverberi et al. 2012  26 NWK No 
Smith, Vartanian, & Goel 2014  16 NWK Yes 
Brunetti et al. 2014  13 NWK Yes 
Porcaro et al. 2014  13 WKA Yes 
Smith, Balkwill, Vartanian, & Goel 2015  17 WKA Yes 

Some data sets contain both WKA and NWK, as well as content and abstract tasks, as different tasks were tested during the experiment.

Baseline Contrasts

The results from the selected experiments are based on a previously chosen contrast between two tasks. As the brain activation pattern accompanying the baseline condition is subtracted from the one during the reasoning condition, the choice of baseline is of crucial importance to the interpretation of subsequent results. There has been a recent controversy about inappropriate baseline conditions in reasoning studies (Prado, 2018; Monti & Osherson, 2012). Baseline conditions can either be chosen too broadly (e.g., fixation cross), leading to results that do not only reflect the reasoning process per se but also ancillary cognitive processes, such as reading. Conversely, they can be too strict (e.g., easy reasoning tasks), resulting in subtraction of crucial aspects of the reasoning process. During study selection, we considered these aspects and only included studies with appropriate baselines. Additionally, we have retrospectively analyzed the baseline conditions, with regard to the studies contributing to the significant activation clusters found in the main analyses. For this procedure, we have excluded the data by Ermer, Guerin, Cosmides, Tooby, and Miller (2006), Fiddick, Spampinato, and Grafman (2005), Houdé et al. (2000), Knauff, Mulack, Kassubek, Salih, and Greenlee (2002), and Van Hoeck et al. (2014), which featured baseline conditions that did not ensure the subtraction of reading from the reasoning condition.

Syllogistic and Conditional Reasoning

Regarding the study by Prado et al. (2011), we evaluated the tasks declared as “categorical” as syllogistic and the “propositional” tasks as conditional but analyzed each study individually to ensure proper task categorization. During the subsequent online research, we explicitly searched for syllogistic and conditional tasks and determined them by analyzing the reasoning process they require. There were exceptions from the abovementioned standard examples of syllogistic and conditional tasks, respectively, but we decided to emphasize the importance of more relaxed inclusion criteria for the sake of obtaining more robust results (see Tables 2 and 3).

Table 3. 
Details of the Article Categorizations Regarding the Quantitative Group Parameters
TypeCategoryNumber of ExperimentsNumber of SubjectsActivation Foci
Reasoning (conditional, syllogistic) All 32 524 549 
 Content 18 256 273 
 Abstract 16 293 263 
  
Conditional All 15 269 353 
 Content 78 145 
 Abstract 10 191 208 
  
Syllogistic All 17 255 196 
 Content 13 178 128 
 Abstract 102 55 
 WKA 110 69 
 NWK 11 170 114 
TypeCategoryNumber of ExperimentsNumber of SubjectsActivation Foci
Reasoning (conditional, syllogistic) All 32 524 549 
 Content 18 256 273 
 Abstract 16 293 263 
  
Conditional All 15 269 353 
 Content 78 145 
 Abstract 10 191 208 
  
Syllogistic All 17 255 196 
 Content 13 178 128 
 Abstract 102 55 
 WKA 110 69 
 NWK 11 170 114 

Concerning the types “Reasoning” and “Syllogistic,” the added world knowledge/content experiments appear to be larger than the total count. This is because, in the experiments by Goel et al. (2000) and Goel and Dolan (2003), both WKA/NWK and content/abstract tasks were tested, and the results were split per category.

For conditional reasoning, these include the experiment by Monti et al. (2007), testing problems such as “If X is either Y or Z, then A.” Furthermore, there were more complex multistep conditional tasks, delivered by Prado and Noveck (2007), Monti et al. (2009), and Reverberi et al. (2010), which we included despite a higher complexity in terms of cognitive load. For syllogistic experiments, these include the study by Goel et al. (1997) in which half of tested tasks were classical syllogisms, whereas the other tasks consisted of other deductive reasoning tasks. We decided to include this study because the measured brain activation reported in the study (at least partly) reflects syllogistic reasoning. The tasks by Kroger et al. (2008) were insofar unusual as they included concrete quantities of items instead of the quantifier “some.” We concluded that the task setup is still a modified form of syllogistic reasoning.

Content and World Knowledge

A common distinction in reasoning tasks is between different types of content (Goel, 2007). These can either be abstract, like letters or symbols, or they can bear semantic content. Hence, we divided the experiments into two further categories: whether the tasks deal with contents referring to any concrete scenario or whether they are abstract in nature (e.g., geometrical figures, number, letters, or nonsensical words). Examples for this can be extracted from the article by Goel et al. (2000), as here, both content and abstract tasks are tested. As content tasks, they use problems like “All poodles are pets. All pets have names. All poodles have names.” and “All P are B. All B are C. All P are C.” as abstract problems.

Especially in the case of tasks including semantic contents, the contents can interfere with the reasoning process by delivering contradictory information regarding the logically correct answer and thereby defeating it (Goel, 2007; Evans, Barston, & Pollard, 1983). This line was also drawn by Fodor (1983) by referring to modular and central cognitive processes, respectively, for processes requiring extensive amounts of world knowledge and those which do not (Oaksford, 2015). To account for these effects, we categorized both syllogistic and conditionals tasks in terms of whether participants had to apply world knowledge potentially affecting the task-solving process by interference with the logical outcome (Hagoort et al., 2004). World knowledge application (WKA) tasks required the application of world knowledge, whereas no world knowledge (NWK) tasks did not and were solved based on the premises' structure at hand. This implies that, for example, belief biases could be only applied to WKA but not to NWK tasks, as only WKA tasks would be considered enthymematic reasoning, that is, reasoning that draws upon world knowledge to use as additional information in problem-solving (Oaksford, 2015; Dennett, 1998). This is because NWK tasks do not contain semantic ambiguities. The tasks by Goel et al. (2000) exemplify our categorization. In a task featuring a belief bias, the participants' conclusion based on his or her world knowledge is opposed to the conclusion inferred by logical thinking. An example would be the following: “All pets are poodles. All poodles are vicious. All pets are vicious.” Because it is generally assumed that pets are not vicious, the (logically) correct conclusion (“All pets are vicious”) is not the same as the conclusion derived by world knowledge (“Not all pets are vicious”). Opposed to that, Goel et al. (2000) also tested abstract tasks such as “All P are B. All B are C. All P are C.” in which no reference to the external world is made; hence, the task does not carry an ambiguous meaning. Nonetheless, we could only find syllogistic tasks that could be characterized as WKA.

Activation Likelihood Activation

For conducting the meta-analysis of neuroimaging data, we chose to use the activation likelihood estimation (ALE) method (Eickhoff, Bzdok, Laird, Kurth, & Fox, 2012; Turkeltaub et al., 2012; Eickhoff et al., 2009), as it is particularly suitable for conducting meta-analyses (for recent examples, see Cona & Scarpazza, 2019; Wertheim & Ragni, 2018; Hobeika, Diard-Detoeuf, Garcin, Levy, & Volle, 2016). By using ALE (GingerALE Version 3.0, www.brainmap.org/ale/), we determine which brain regions exhibit activation during the cognitive processing of a specific task. These activation loci are based upon the data set of stereotactic 3-D coordinates retrieved from neuroimaging studies. These represent the changes in BOLD response between task conditions. Because most neuroimaging studies only involve around 10–20 participants, it is particularly interesting to conduct meta-analyses featuring several experiments. This cross-study approach is facilitated by ALE, yielding results across differences in experimental setups and testing materials. GingerALE offers two options for analyzing data sets: (1) analyzing a single data set and (2) comparing and contrasting two different sets (Eickhoff et al., 2011). We used both methods to produce detailed results.

The maximum activation likelihood is generated by rendering modeled activation (MA) maps from the results of each experiment included in the data set. The MA maps are reconstructed maps of brain activation collected by scanning data. Reconstruction is achieved by treating each data point as the center of a Gaussian distribution, which is blurred by the FWHM determined by the number of participants (Eickhoff et al., 2009). In the next step, the MA maps are combined and the likelihood of activation in each voxel of the accumulated MA maps is calculated, regardless of the respective location. The MA maps, the p value image, and the threshold are combined for the thresholded ALE map (Eickhoff et al., 2009, 2012). Eickhoff et al. (2016) recommend the use of the cluster-level family-wise error method in which a randomized data set tantamount to the testing data set is generated. The sets are comparable with regard to subject size, number of foci, and number of studies and are subsequently compared.

ALE Parameters

After having retrieved the activation coordinates from the respective experiments, we converted foci represented in Talairach space to MNI space by the conversion algorithm “icbm2tal” (Laird et al., 2010) to enable consistent data processing. In the single-study analyses, as recommended by Eickhoff et al. (2016), we evaluated the data by cluster-level analyses with a p value of .01 and 1000 threshold permutations with the nonparametric statistic method (Turkeltaub et al., 2012). The prior uncorrected p value is .001. In the contrast and conjunction studies, we used the previously mentioned parameter for analyzing the single groups. For group comparison, we applied an uncorrected p value of .01 and 10,000 permutations. We excluded all results with a volume size smaller than 200 mm2.

RESULTS

In this study, we analyzed 32 neuroimaging experiments to identify the neural correlates of conditional and syllogistic reasoning (Figure 1; Table 3), as well as the dissociations in terms of brain activation between different subdivisions, such as requirement of content and world knowledge (for a schematic overview, see Table 4; for detailed results, see Appendix). In the following, we are going to report the results in terms of anatomical location, as well as Brodmann's area (BA).

Figure 1. 

From left to right, top to bottom: 1–3: conditional reasoning; 4–6: conditional reasoning, content; 7–9: conditional reasoning, abstract, 10–12: syllogistic reasoning; 13–15: syllogistic reasoning, content; 16–18: syllogistic reasoning, abstract; 19–21: syllogistic reasoning, WKA; 22–24: syllogistic reasoning; NWK; 25–27: content tasks; 28–30: abstract tasks.

Figure 1. 

From left to right, top to bottom: 1–3: conditional reasoning; 4–6: conditional reasoning, content; 7–9: conditional reasoning, abstract, 10–12: syllogistic reasoning; 13–15: syllogistic reasoning, content; 16–18: syllogistic reasoning, abstract; 19–21: syllogistic reasoning, WKA; 22–24: syllogistic reasoning; NWK; 25–27: content tasks; 28–30: abstract tasks.

Table 4. 
Schematic Representation of the Significant Activation Clusters of the ALE Analyses
TypeLobeFrontalParietalO.S.Pos.
Brodmann Area10474546944864073919181732CAUCT
All All ◖ ◖   ◐ ◐   ◖ ◖ ◖   ◖ ◖ ◖ ◗ ◖ ◗   
Content             ◖ ◖ ◖ ◖     ◖   ◖     
Abstract ◖ ◖   ◖ ◖ ◖ ◖ ◖ ◖   ◖ ◖ ◖         
Conditional \ Syllogistic             ◖ ◖ ◖       ◖ ◖       
Content ∪ Abstract             ◖   ◖       ◖         
Content \ Abstract                         ◖ ◗       
  
Conditional All ◖ ◖     ◖   ◖ ◖ ◖   ◖   ◖       ◗ 
Content         ◖   ◖ ◖         ◖         
Abstract ◖ ◖       ◖ ◖ ◖ ◖ ◖ ◖   ◖       ◗ 
Content ∪ Abstract             ◖ ◖         ◖         
  
Syllogistic All     ◖ ◖ ◖               ◖   ◖     
Content           ◖             ◖         
Abstract   ◖   ◖ ◖     ◖   ◖     ◖         
Content ∪ Abstract         ◖ ◖             ◖         
Abstract \ Content                   ◖     ◖         
WKA           ◖             ◖         
NWK   ◖   ◖ ◖               ◖         
WKA ∪ NWK         ◖ ◖             ◖         
TypeLobeFrontalParietalO.S.Pos.
Brodmann Area10474546944864073919181732CAUCT
All All ◖ ◖   ◐ ◐   ◖ ◖ ◖   ◖ ◖ ◖ ◗ ◖ ◗   
Content             ◖ ◖ ◖ ◖     ◖   ◖     
Abstract ◖ ◖   ◖ ◖ ◖ ◖ ◖ ◖   ◖ ◖ ◖         
Conditional \ Syllogistic             ◖ ◖ ◖       ◖ ◖       
Content ∪ Abstract             ◖   ◖       ◖         
Content \ Abstract                         ◖ ◗       
  
Conditional All ◖ ◖     ◖   ◖ ◖ ◖   ◖   ◖       ◗ 
Content         ◖   ◖ ◖         ◖         
Abstract ◖ ◖       ◖ ◖ ◖ ◖ ◖ ◖   ◖       ◗ 
Content ∪ Abstract             ◖ ◖         ◖         
  
Syllogistic All     ◖ ◖ ◖               ◖   ◖     
Content           ◖             ◖         
Abstract   ◖   ◖ ◖     ◖   ◖     ◖         
Content ∪ Abstract         ◖ ◖             ◖         
Abstract \ Content                   ◖     ◖         
WKA           ◖             ◖         
NWK   ◖   ◖ ◖               ◖         
WKA ∪ NWK         ◖ ◖             ◖         

The semicircles indicate whether a significant activation cluster was found in the respective cerebral hemisphere. The semicircles indicate the laterality of activation in the respective BA. If both semicircles are indicated, the filled one indicates the side in which the larger cluster was found. O. = occipital; S. = sublobar; Pos. = posterior.

Single Studies

During general reasoning, that is, syllogistic, as well as conditional reasoning, we found a wide-spread activation network encompassing the frontal, parietal, sublobar, limbic, posterior lobes. Concerning the frontal lobe, we found activation in the bilateral IFG (left: BA 9, BA 47, BA 10; right: BA 9), bilateral middle frontal gyrus (MFG; left: BA 8, BA 6, BA 46, BA 47, BA 10; right: BA 9, BA 46), left PreCG (BA 9, BA 6), left superior frontal gyrus (SFG; BA 6, BA 10), left medial frontal gyrus (MeFG, BA 8, BA 6), and BA 47. Concerning the parietal lobes, we found left-sided activity in the inferior parietal lobule (IPL; BA 40), the angular gyrus (ANG; BA 39), and the precuneus (PCUN; BA 19). We identified sublobar and limbic activation in the left caudate (CAU) and right cingulate gyrus (CG; BA 32). Further clusters were found in the right cerebellar tonsil (CT).

During content tasks, left-sided activation was found in the MFG (BA 6), MeFG (BA 8), SFG (BA 6, BA 8), IPL (BA 40), superior parietal lobule (SPL; BA 7), CAU, and lentiform nucleus. Right-sided activation was found in the lingual gyrus (LiG; BA 17) and the inferior occipital gyrus (IOG; BA 18). During abstract tasks, activation was exclusively in the left-sided parietal lobe in the left IPL (BA 40) and PCUN (BA 19, BA 39) and the left frontal lobe at the IFG (BA 9, BA 44, BA 47), MFG (BA 46, BA 6, BA 47, BA 10), SFG (BA 6), MeFG (BA 8), PreCG (BA 6), and BA 47.

Analyzing all 15 conditional experiments, we identified bilateral frontal, parietal, and cerebellar activation. These include the left MeFG (BA 8, BA 6), SFG (BA 6), MFG (BA 8, BA 6, BA 47), PreCG (BA 9, BA 6), and IFG (BA 19). The parietal areas encompass the left IPL (BA 40) and ANG (BA 39). Further areas include the left BA 47 and the right inferior semilunar lobule and CT. For the five conditional content experiments, activation was found in the left MeFG (BA 8, BA 9, BA 6), left SFG (BA 6), and MFG (BA 6, BA 8). The 10 abstract conditional experiments yielded left-lateralized activation in parietal and frontal cortices as well as in right-sided posterior areas, including the IPL (BA 40), PCUN (BA 39), SPL (BA 7), MFG (BA 10, BA 47, BA 6), IFG (BA 10, BA 6, BA 44), SFG (BA 10), MeFG (BA 8), left BA 47, as well as the right-sided CT, tuber, and inferior semilunar lobule.

From analyzing 17 experiments testing syllogistic reasoning, activation in the IFG (BA 9, BA 45), MFG (BA 46), as well as the CAU was found. For the 13 syllogistic content tasks, activation was localized in the left PreCG (BA 44) and IFG (BA 44), whereas the six syllogistic abstract tasks elicit activation in the left IFG (BA 9, BA 47), MFG (BA 46, BA 6), and PCUN (BA 7). Concerning the eight syllogistic WKA tasks, we found exclusively left-sided activation in the IFG (BA 44). During the 11 syllogistic NWK tasks, activation was exhibited in the left-sided IFG (BA 47, BA 9), as well as MFG (BA 46).

Contrast Studies

In contrast to syllogistic tasks, areas exclusively active during conditional encompass the left-sided IPL (BA 40), left MeFG (BA 8, BA 6), left MFG (BA 6, BA 8), and left CG (BA 32). Areas commonly activated by content and abstract tasks encompass the left IPL (BA 40) and MeFG (BA 8), whereas content tasks exclusively activate the right LiG (BA 17). The active area that conditional content and abstract tasks share are the left MeFG (BA 8, BA 6). Syllogistic content and abstract tasks share activation in the left IFG (BA 44, BA 9). The contrast analysis between syllogistic content and abstract tasks reveal that the abstract tasks are accompanied by activation in the left PCUN (BA 7). Regarding syllogistic tasks, we conducted a contrast and conjunction analysis on WKA and NWK tasks. Here, we found shared areas in the left IFG (BA 44, BA 9; Table 4).

DISCUSSION

In the previous chapters, we described how we selected 32 reasoning experiments and categorized them based on their design and the cognitive capabilities their task setup requires. Also, we excluded inappropriate baseline conditions that might influence the subsequent results by ensuring that the results only reflect the reasoning process itself but not axillary processes. Based on this selection, ALE analyses on single data sets, as well as contrast and conjunction analyses, were conducted to identify the brain regions active during the tasks at hand. In the following, we are going to conceptualize the results and base the interpretation upon the findings from cognitive neuroscience, psychology, and modeling research. In particular, we focus on the predictions of cognitive theories providing algorithmic descriptions of the reasoning process from which the implementation in terms of brain activation can be inferred. This is not to say that cognitive theories of reasoning predict brain activation per se, but that they predict more elementary cognitive processes that are more closely associated with activation in a particular brain region. By the identification and localization of these cognitive processes, we aim at closing the symbolic–subsymbolic gap (Goertzel, 2012) between high-level and low-level cognitive processes. We have summarized the predictions concerning brain activation for each previously described theory in Table 6. We have to note that neither theory is associated with detailed descriptions of brain activation during the reasoning process. Especially for dual process accounts of reasoning, it is difficult to infer brain activation patterns, as the exact cognitive processes involved in both systems are not thoroughly described and hence contains many degrees of freedom (Keren, 2013). Nonetheless, they still serve as a theoretical scaffolding for interpreting the subsequent results and hence allow to interpret the results beyond dissociation (Henson, 2006).

What Are the Neural Correlates of Conditional Reasoning?

During the processing of conditional reasoning tasks, we found a widespread activation network, including the frontal, parietal, and posterior lobes. Herewith, we have replicated the results by Prado et al. (2011) and additionally identified regions in the frontal and parietal cortices, as well as the cerebellum (see Table 5). Based on our further findings, we can conclude that the results from Prado et al. (2011), mostly located in prefrontal areas, could be further differentiated by the content differentiation (see Table 4).

Table 5. 
Simplified Representation of the Significant Activation Clusters Found by This Study and Prado et al. (2011)
graphic
graphic

The semicircles indicate whether a significant cluster was found in the respective cerebral hemisphere. The semicircles indicate the laterality of activation in the respective BA. Blue color stands for results from this study, red for the results by Prado et al. (2011), violet for areas reported in both studies. O = occipital; Pos. = posterior; Pu. = putamen.

Table 6. 
Predictions of Brain Activation Patterns by the Currently Prevalent Cognitive Theories of Reasoning
TheoryPrediction(s)
Mental model theory Right hemisphere (Knauff et al., 2003; Parsons & Osherson, 2001), potentially parietal cortex (Goel et al., 2000
Mental logic theory Left hemisphere (Knauff et al., 2003), language-related areas (Parsons & Osherson, 2001; Goel et al., 2000
Dual process theory Two processes, dependent on the task's contents 
System 1: frontotemporal pathway 
System 2: parietal pathway (Goel, 2003
Bayesian probabilistic account No predictions in terms of brain activation patterns (Oaksford, 2015
TheoryPrediction(s)
Mental model theory Right hemisphere (Knauff et al., 2003; Parsons & Osherson, 2001), potentially parietal cortex (Goel et al., 2000
Mental logic theory Left hemisphere (Knauff et al., 2003), language-related areas (Parsons & Osherson, 2001; Goel et al., 2000
Dual process theory Two processes, dependent on the task's contents 
System 1: frontotemporal pathway 
System 2: parietal pathway (Goel, 2003
Bayesian probabilistic account No predictions in terms of brain activation patterns (Oaksford, 2015

The largest activation cluster accompanying conditional reasoning encompasses the left IPL and ANG, supporting the findings by Prado et al. (2011). In the domain of reasoning, the PPC is associated with the construction of the internal representation of the premises' contents, as well as more generally visuospatial working memory (Mellet, Petit, Mazoyer, Denis, & Tzourio, 1998). The pronounced activation in this region could support a mental model account of reasoning for conditional tasks.

Concerning the frontal cortex, we found activation in the BA 8 and BA 10, which were proposed to be the central regions underlying deductive reasoning (Monti & Osherson, 2012) and identified to be responsible for facilitating the guiding of mental operations performed on the premises to reach a conclusion by managing operations and the allocation of cognitive resources (van den Heuvel et al., 2003). BA 8 allows to monitor cognitive subtasks (Koechlin, Corrado, Pietrini, & Grafman, 2000) and rule-based information gaiting (Volz, Schubotz, & von Cramon, 2005). A lesion study by Reverberi, Shallice, D'Agostini, Skrap, and Bonatti (2009), investigating left and right, and medial prefrontal lesions with respect to solving deductive reasoning tasks support this claim by stating that the medial portion of the pFC is crucial for identification and maintenance of the overall problem structure. Furthermore, the MeFG is active during abstract rule maintenance (Bunge, Kahn, Wallis, Miller, & Wagner, 2003), as well as executive control (Posner & Dehaene, 1994) and mental goal coordination (Fletcher & Henson, 2001). The dorsolateral pFC (DLPFC) has been associated with evaluation and manipulation of information present in working memory (Christoff et al., 2001; Fletcher & Henson, 2001), as well as integration of chunks of information (De Pisapia & Braver, 2008; Kroger et al., 2002; Petrides, 1995). The frontal activation of the DLPFC, as well as the SMA, could be conceptualized with regard to the cerebellar activation found in the CT. Cerebellar activation had not yet been largely regarded in the interpretation of neuroimaging studies of reasoning. In a neuroimaging study from 2004, Blackwood et al. found the coactivation of the cerebellum with the SMA, IPL, and occipital cortex to mediate decision-making under uncertainty. Moreover, this network was found active during conceptual reasoning (Rao et al., 1997). It was suggested that the cerebellum plays a key role in the construction of a mental working model of the world under uncertainty (Blakemore, Frith, & Wolpert, 2001; Ito, 1993).

In conditional reasoning, a mental model of the hypothetical situation described in the premises has to be constructed, as expressed by the antecedent clause “if….” Subsequently, this hypothetical model has to be examined based on the consequent's assumptions expressed by the statement “…then.” Based on a mental model framework, this task could be accomplished by a model of the external world mediated by parietal activation for model maintenance (Knauff et al., 2002) and BA 10 and BA 8 activation subserving the gating of information processing. Nonetheless, this interpretation is only speculative. Furthermore, it is assumed that the cerebellar–thalamic–DLPFC network serves as a mechanism for integrating and gating of structured thought (Schmahmann, 1991, 1996). Interestingly, Noveck et al. (2004) found that the solving of Modus Ponens tasks evoked activity in the left PPC, whereas the processing of MT tasks is accompanied by activation in the left IFG. This could be explained by the differences in the second premise these tasks posit. During MT tasks, the second premise (p) is an affirmative hypothetical statement that might be easier to solve via a mental model approach, whereas MT tasks involve a negative statement (not-q), the integration of a negative statement for which the mental model needs to be varied. Concerning the content classification of conditional tasks, we found a shared activation cluster for both task types recruiting the MeFG (BA 8). Hence, this region is involved in conditional reasoning regardless of the tasks' specific contents. Content task solving evokes activation in the left SMA and DLPFC, whereas abstract task solving recruited a more wide-spread network, also encompassing the left PPC, left rostrolateral PFC (RLPFC), IFG, SMA, and right cerebellum. As especially abstract tasks involve activation in the PPC, we would rather assume a model-based reasoning in abstract tasks, whereas content-based tasks might elicit processes that are rather focused on the real-world relevance conveyed by the premises' contents.

What Are the Neural Correlates of Syllogistic Reasoning?

Left-sided inferior and middle frontal (BA 9, BA 45) and CAU activation was identified during syllogistic reasoning. As opposed to Prado et al. (2011), we did not find any activation in the right BG and left PreCG, but newly identified activation in the left MFG (BA 46; Table 5). Concerning the frontal regions, we found the largest activation cluster in the left IFG/MFG, hence Broca's area (Ardila, Bernal, & Rosselli, 2016). Broca's area is known to be involved in language processing (Friederici, 2011; Grodzinsky & Santi, 2008; Friederici & Kotz, 2003; Pickett, Kuniholm, Protopapas, Friedman, & Lieberman, 1998). It is a syntax-specific area (Embick, Marantz, Miyashita, O'Neil, & Sakai, 2000) involved in inspection of syntactic anomaly (Ni et al., 2000) and correct grammatical constructions and specifically syntactic movement (Grodzinsky & Santi, 2008).

Furthermore, BA 45 was found to be involved in the formation of semantic relationships (Friederici, 2002), as well as an area active during the processing of especially complex syntactic structures (Friederici, Fiebach, Schlesewsky, Bornkessel, & von Cramon, 2006; Grodzinsky & Friederici, 2006; Newman, Just, Keller, Roth, & Carpenter, 2003). The left IFG is active during syntactic structure building processes (Meyer, Friederici, & von Cramon, 2000), indicating that syllogistic reasoning might be based on the formation and evaluation of syntactical constructions as the means for problem-solving. Generally, the IFG is associated with early building-type syntactic processes, whereas the BG are associated with later syntactic assessment processes (Friederici & Kotz, 2003). Also, the activation of left lateral BG activity is coherent with the interpretation of Prado et al. (2011), who claimed that this region is commonly coactivated with the left IFG in tasks involving syntactic processing (Friederici & Kotz, 2003, Friederici, Kotz, Werheid, Hein, & von Cramon, 2003; Moro et al., 2001; Ni et al., 2000). Anatomically, the BG project to the frontal cortex via the thalamus (Ullman, 2006). Two commonly assumed functions are the support of procedural memory and declarative knowledge retrieval, such as implemented in the cognitive architecture ‘Adaptive Control of Thought-Rational’ (ACT-R; Anderson, 2007). The left-sided CAU nucleus is involved in syntactic processes (Ni et al., 2000), more specifically in the later stages of syntax integration and evaluation (Friederici & Kotz, 2003). A lesion study by Hochstadt, Nakano, Lieberman, and Friedman (2006) exhibited that degeneration of the BG inhibits syntactic ability, even without interfering with working memory capacity. Regarding the interaction between BG and IFG, Ullman (2006) suggested that the frontostriatal system facilitates the formation of syntactic structures, such as sentences from smaller units by recursive rules such as syntax. Also, Friederici et al. (2006) stated that the specific involvement of the BG stems from their facilitation of content reconsideration when syntactic structures fail to provide a coherent pattern, which is additionally supported by a lesion study (Kotz, Frisch, von Cramon, & Friederici, 2003; for a review, see Tettamanti et al., 2005).

It is notable that we did not find any parietal activation, as opposed to conditional and relational tasks (Wertheim & Ragni, 2018; Prado et al., 2011). This supports the claim that syllogistic premises are not represented in a classic mental model scheme. This is consistent with the conclusions drawn by Prado et al. (2011) in which they reject the idea of visuospatial premise processing in categorical tasks. They argue that categoricals pose an enhanced difficulty of visuospatial mental representation (Favrel & Barrouillet, 2000). Theoretically, our findings support this assumption. Hence, even if the arguments do not carry world knowledge and therefore context, syllogisms are not likely to be represented as mental models but are accompanied by activation in brain regions associated with syntactic capabilities. It might be argued that especially because syllogisms are hard to envision mentally (because they convey category membership, which can be represented in multiple ways), participants switch approaches and rely on the syntactic structure of the premises (Prado et al., 2011). Neither the left BA 10 nor the left BA 8 was found reliably active across syllogistic task results. Hence, it is debatable whether the two regions can be defined as the core region for deductive reasoning. As we have controlled for baseline conditions that might yield residual activation evoked by reading, the language-related activation during syllogistic reasoning cannot be reduced to inappropriate task design but is rather as inherently evoked by the mental processing of syllogisms themselves.

Concerning the world knowledge differentiation in syllogistic tasks, differences in prefrontal activation between the two task types were observed. WKA tasks evoke activation in the left BA 44 (PreCG and IFG), whereas NWK tasks involve BA 9, BA 47 (IFG), and BA 46 (MFG) activity. The contrast analyses between those two types reveal that the shared area between those two task types encompasses the left IFG (BA 44, BA 9). These areas can be considered part of “Broca's complex,” a recently defined cluster of prefrontal BAs and the BG (Ardila et al., 2016), which are found to be associated with language production (Ardila, Bernal, & Rosselli, 2017; Hagoort, 2005; Lemaire et al., 2013) and syntax processing (Friederici et al., 2006). Monti and Osherson (2012) address this finding from the experiment by Goel et al. (2000), suggesting that the activation in this region (between tasks with and without semantically meaningful content) may be due to an inaccurate baseline condition, not controlling for brain activity accompanying reading. As we have controlled for baseline conditions, our finding strengthens the result by Goel et al. (2000) for BA 44 being a region central to both content types. Interestingly, the content/abstract differentiation of syllogistic tasks reveals a very similar pattern with IFG (BA 44) activation during content tasks and activation in the left IFG/MFG (BA 9, BA 46, BA 47, BA 6) and PCUN (BA 7) during abstract tasks. The shared area encompasses the left IFG (BA 44, BA 9). Activation in the left PCUN is exclusively accompanying the processing of abstract tasks. Hence, we assume that the world knowledge and content differentiations, which are justified by (1) the information given to the reasoner (content) and (2) the information needed to successfully solve the task (world knowledge), do not meaningfully reflect an underlying processing difference in terms of cognitive mechanisms.

It appears to be that syllogistic reasoning is only accompanied by activation in Broca's complex. This poses the question of what the process of reasoning comprises in the case of syllogistic reasoning. This is especially interesting with regard to relational reasoning. Here, we find activation in the frontoparietal network and associated mental model construction and manipulation (Wertheim & Ragni, 2018). As in syllogistic reasoning, only areas usually known to be involved in language production are active, the solving of syllogistic tasks seems to be best described by the tenets of mental logic accounts of reasoning. This would imply that syllogistic statements are processed by mainly syntax manipulation instead of mental model construction. Nonetheless, it might also be accounted for by a heuristic approach to task solving, as can be inferred from the diverging brain activation patterns in WKA/NWK and content/abstract tasks.

The single-study analyses indicate that syllogistic and conditional reasoning elicit fundamentally different brain activation patterns. Furthermore, the conjunction analysis between the two data sets did not yield any significant overlap in terms of brain activation, further strengthening this assumption. Contrast analysis revealed that conditional reasoning, in contrast to syllogistic reasoning, elicits activation in a widespread network encompassing the left IPL (BA 40), as well as the frontal MeFG and MFG (BA 8, BA 32, BA 6), as expected from the single-study analysis. As these regions are especially active during conditional in contrast to syllogistic reasoning, this supports the claim that crucially different processes underlie task solving either problem type. As previously mentioned, the PPC and BA 8 might support the reasoning process by facilitating the mental representation of the situation depicted in the premises. Furthermore, BA 8 served to monitor subtasks needed to successfully solve the task at hand. These regions not being found reliably active during syllogistic reasoning does not imply that these processes do not subserve the process. Nonetheless, it shows the neurological dissociation between the solving of both task types, indicating distinct underlying processes. For syllogistic reasoning, we assume that participants rather focus on the structure of the premises instead of on building a mental model of the situation presented. For conditional reasoning, the additionally active regions indicate more involvement of regions needed for mental representation.

What Are the Differences in Neural Activation during Content and Abstract Tasks?

The analyses of the data sets with respect to the tasks' contents shows that the brain activation pattern of content tasks includes the left SMA (BA 6) and MeFG (BA 8), the left PPC, the right LiG and IOG (BA 17, BA 18), as well as the left BG. Abstract tasks activate the left-sided Broca's area (IFG/MFG), the parietotemporal junction, and the RLPFC (BA 46, BA 10). Relating back to the theoretical frameworks, dual process theory could be considered to explain these patterns. This is especially the case, as we did find activation in the parietal and frontotemporal networks in content and abstract tasks, respectively. Surprisingly, these have been associated by Goel (2003) as the activation patterns associated with System II and I processing, respectively. Coming from a processing perspective, we would rather assume the observed brain activation patterns to accompany the converse content type. Nonetheless, the striking differences in brain activation between content and abstract tasks might still be compatible with a dual process account of reasoning. Hence, it is necessary to more deeply investigate the algorithmic levels (Marr, 1982) underlying Systems I and II, respectively, to infer more sophisticated descriptions of the cognitive processes and hence brain activation patterns involved.

The conjunction and contrast analyses show that both groups (content and abstract) exhibit activation overlap in the left IPL and MeFG (BA 8). Reverberi et al. (2012) identified predictive brain activation profiles for syllogistic reasoning and proposed that activation in, inter alia, the SFG (BA 6, BA 8) serves as a predictor for participants' identification of logical structures. Hence, logical analysis can be assumed regardless of the necessity for content application.

Conclusion

After having collected and statistically analyzed neuroimaging studies investigating human reasoning, we elucidated the neural and cognitive mechanisms accompanying conditional and syllogistic task solving. We found that the brain activation patterns underlying conditional and syllogistic reasoning consistently differ. This challenges the assumption that there is one brain network and cognitive process subserving all types of task solving subsumed under the category of reasoning. Rather, we propose that reasoning is facilitated by at least two different networks, either mental model-based or syntax-based, triggered by the task and content type. To meaningfully interpret the results, different conceptual frameworks were applied. Regarding the processing of conditional problems, we found activation in the left IPL, DLPFC, RLPFC, and SMA, subserving working memory capacity and information integration, as well as coactivation of cerebellar regions. This could be explained by the hypothetical structure of conditional tasks, which necessitate the formation and maintenance of a mental model subserved by the frontoparietal network. Syllogistic reasoning elicits activation in the left IFG/MFG, hence Broca's area, coactive with the BG. Interestingly, parietal as well as cerebellar activities were absent, which we interpret as indicative for syntax-based processing. Concerning the content differentiation, abstract tasks elicit activation in the left MFG, RLPFC, SMA, as well as IPL, whereas content tasks are subserved by the left MFG, BG, and LiG. Nonetheless, both divisions share a great overlap in the left SFG, indicating the analysis of the underlying logical structure regardless of the necessity for content application.

At large, these findings rule out theoretical frameworks relying on single approaches, such as mental model theory and mental logic theory as a mean to explain the cognitive processing of all types of reasoning tasks. Also, Bayesian approaches do not seem to adequately describe the neurological aspects of the reasoning process, as different task types do evoke different neural pathways. Instead, there seem to be different strategies to solve the problems evoked by their structure and content, giving validity to dual process accounts of reasoning. This meta-analysis could shed a new light long-discussed problems in the conceptualization of reasoning. Future work should focus on the more in-depth differentiation between syntactic processes and reasoning, especially in the case of syllogistic reasoning. Furthermore, a more fine-grained temporal resolution of the processes involved in reasoning could reveal new insights about potentially shared and dissociate cognitive mechanisms underlying different task types and foster the understanding of diverging brain activation patterns.

APPENDIX: ALE RESULTS OF THE SINGLE AND CONTRAST ANALYSES

Reasoning TypeData SetCluster Size (mm3)Peak Coordinates (MNI)Hem.LobeAssigned LocationBrodmann Area
xyz
Conditional and syllogistic All 5088 −44 16 27 F, S IFG, MFG, Extranuclear, PreCG 9, 8, 47, 6, 8, 46 
6600 −3 26 47 L, R F, L SFG, MeFG, CG 6, 8, 32 
6464 −43 −55 48 IPL, ANG, PCUN 40, 39, 19 
3024 −44 50 −9 IFG, MFG, SFG 10, 47 
1624 36 −69 −35 Pos CT n/a 
1592 −13 CAU n/a 
1112 47 26 26 MFG 46, 9 
Content 1560 −42 17 43 MFG 
1152 −48 −51 49 IPL, SPL 40, 7 
1128 26 −91 −3 LiG, IOG 17, 18 
1096 −5 35 48 MeFG, SFG 8, 6 
856 −13 CAU, lentiform nucleus n/a 
Abstract 4408 −41 −56 48 IPL, PCUN 40, 19, 39 
3480 −49 15 21 IFG, MFG 9, 46, 44 
2512 −5 23 47 SFG, MeFG 6, 8 
2280 −41 51 MFG, PreCG 
2240 −43 26 −5 F, S IFG, extranuclear, MFG 47 
1904 −43 51 −7 MFG 46, 10 
Syllogistic ∪ Conditional None               
Syllogistic \ Conditional None               
Conditional \ Syllogistic 4312 −41 −56 46 IPL 40 
1984 −5 35 38 F, L MeFG, CG 8, 32, 6 
216 −46 16 47 MFG 6, 8 
Content ∪ Abstract 584 −46 −53 48 IPL 40 
248 −6 35 44 MeFG 8, 6 
Content \ Abstract 208 17 −93   LiG 17 
Abstract \ Content None               
Conditional All 6664 −42 −56 47 IPL, ANG 40, 39 
4456 −4 32 43 MeFG, SFG 8, 6 
4080 −42 16 45 MFG. PreCG 8, 9, 6 
2088 −46 48 −13 MFG, IFG 47, 10 
1448 −39 23 −7 Extranuclear 47 
1224 36 −70 −36 Pos Inferior semilunar lobule, CT n/a 
Content 1728 −5 36 42 MeFG, SFG 8, 6, 9 
1504 −42 17 42 MFG 6, 8 
Abstract 5184 −42 −56 48 IPL, PCUN, SPL 40, 39, 7 
2360 −42 51 −9 MFG, IFG, SFG 10, 47 
1360 −50 11 24 IFG 6, 44 
1320 −3 33 42 MeFG 
1240 −39 23 −6 Extranuclear 47 
1208 36 −70 −35 Pos CT, tuber, inferior semilunar lobule n/a 
904 −36 12 53 MFG 
Content ∪ Abstract 528 −5 34 43 MeFG 
Content \ Abstract None               
Abstract \ Content None               
Syllogistic All 5224 −50 19 13 IFG, MFG 9, 45, 46 
1184 −11 CAU n/a 
Content 1704 −52 20 10 PreCG, IFG 44 
Abstract 1960 −48 18 20 IFG, MFG 9, 46 
1072 −48 29 −1 IFG 47 
864 −45 54 MFG 
688 −2 −60 40 PCUN 
Content ∪ Abstract 384 −53 16 20 IFG 44, 9 
Content \ Abstract None               
Abstract \ Content 648 −2 −60 40 PCUN 
WKA 1656 −53 19 14 IFG 44 
NWK 1752 −48 26 −2 IFG 47 
1736 −48 17 21 IFG, MFG 9, 46 
WKA ∪ NWK 424 −53 16 20 IFG 44, 9 
WKA \ NWK None               
NWK \ WKA None               
Reasoning TypeData SetCluster Size (mm3)Peak Coordinates (MNI)Hem.LobeAssigned LocationBrodmann Area
xyz
Conditional and syllogistic All 5088 −44 16 27 F, S IFG, MFG, Extranuclear, PreCG 9, 8, 47, 6, 8, 46 
6600 −3 26 47 L, R F, L SFG, MeFG, CG 6, 8, 32 
6464 −43 −55 48 IPL, ANG, PCUN 40, 39, 19 
3024 −44 50 −9 IFG, MFG, SFG 10, 47 
1624 36 −69 −35 Pos CT n/a 
1592 −13 CAU n/a 
1112 47 26 26 MFG 46, 9 
Content 1560 −42 17 43 MFG 
1152 −48 −51 49 IPL, SPL 40, 7 
1128 26 −91 −3 LiG, IOG 17, 18 
1096 −5 35 48 MeFG, SFG 8, 6 
856 −13 CAU, lentiform nucleus n/a 
Abstract 4408 −41 −56 48 IPL, PCUN 40, 19, 39 
3480 −49 15 21 IFG, MFG 9, 46, 44 
2512 −5 23 47 SFG, MeFG 6, 8 
2280 −41 51 MFG, PreCG 
2240 −43 26 −5 F, S IFG, extranuclear, MFG 47 
1904 −43 51 −7 MFG 46, 10 
Syllogistic ∪ Conditional None               
Syllogistic \ Conditional None               
Conditional \ Syllogistic 4312 −41 −56 46 IPL 40 
1984 −5 35 38 F, L MeFG, CG 8, 32, 6 
216 −46 16 47 MFG 6, 8 
Content ∪ Abstract 584 −46 −53 48 IPL 40 
248 −6 35 44 MeFG 8, 6 
Content \ Abstract 208 17 −93   LiG 17 
Abstract \ Content None               
Conditional All 6664 −42 −56 47 IPL, ANG 40, 39 
4456 −4 32 43 MeFG, SFG 8, 6 
4080 −42 16 45 MFG. PreCG 8, 9, 6 
2088 −46 48 −13 MFG, IFG 47, 10 
1448 −39 23 −7 Extranuclear 47 
1224 36 −70 −36 Pos Inferior semilunar lobule, CT n/a 
Content 1728 −5 36 42 MeFG, SFG 8, 6, 9 
1504 −42 17 42 MFG 6, 8 
Abstract 5184 −42 −56 48 IPL, PCUN, SPL 40, 39, 7 
2360 −42 51 −9 MFG, IFG, SFG 10, 47 
1360 −50 11 24 IFG 6, 44 
1320 −3 33 42 MeFG 
1240 −39 23 −6 Extranuclear 47 
1208 36 −70 −35 Pos CT, tuber, inferior semilunar lobule n/a 
904 −36 12 53 MFG 
Content ∪ Abstract 528 −5 34 43 MeFG 
Content \ Abstract None               
Abstract \ Content None               
Syllogistic All 5224 −50 19 13 IFG, MFG 9, 45, 46 
1184 −11 CAU n/a 
Content 1704 −52 20 10 PreCG, IFG 44 
Abstract 1960 −48 18 20 IFG, MFG 9, 46 
1072 −48 29 −1 IFG 47 
864 −45 54 MFG 
688 −2 −60 40 PCUN 
Content ∪ Abstract 384 −53 16 20 IFG 44, 9 
Content \ Abstract None               
Abstract \ Content 648 −2 −60 40 PCUN 
WKA 1656 −53 19 14 IFG 44 
NWK 1752 −48 26 −2 IFG 47 
1736 −48 17 21 IFG, MFG 9, 46 
WKA ∪ NWK 424 −53 16 20 IFG 44, 9 
WKA \ NWK None               
NWK \ WKA None               

The significant clusters as well as differentiated anatomical localizations are reported. “∪” indicates the conjunction analysis of the two sets; “\” indicates the subtraction analysis of the right from the left set. Hem. = hemisphere; L = left; R = right; F = frontal; S = sublobar; L = limbic; P = parietal; Pos = posterior; O = occipital; n/a = not available.

Acknowledgments

This research was supported by the BrainLinks-BrainTools Cluster of Excellence funded by the German Research Foundation (DFG, grant number EXC 1086) and the Barbara-Wengeler Foundation to J. W., a Heisenberg scholarship (RA 1934/3-1 and RA 1934/4-1) to M. R., and the DFG priority program New Frameworks of Rationality (SPP 1516).

Reprint requests should be sent to Julia Wertheim, Cognitive Computation Lab, Albert-Ludwigs-Universität Freiburg, Georges-Köhler-Allee 79, 79110 Freiburg, Germany, or via e-mail: wertheim@cs.uni-freiburg.de.

REFERENCES

References marked with an asterisk (*) indicate studies included in the meta-analysis.

Anderson
,
J. R.
(
2007
).
How can the human mind occur in the physical universe?
New York
:
Oxford University Press
.
Ardila
,
A.
,
Bernal
,
B.
, &
Rosselli
,
M.
(
2016
).
How localized are language brain areas? A review of Brodmann areas involvement in oral language
.
Archives of Clinical Neuropsychology
,
31
,
112
122
.
Ardila
,
A.
,
Bernal
,
B.
, &
Rosselli
,
M.
(
2017
).
Should Broca's area include Brodmann area 47?
Psicothema
,
29
,
73
77
.
*
Baggio
,
G.
,
Cherubini
,
P.
,
Pischedda
,
D.
,
Blumenthal
,
A.
,
Haynes
,
J.-D.
, &
Reverberi
,
C.
(
2016
).
Multiple neural representations of elementary logical connectives
.
Neuroimage
,
135
,
300
310
.
Blackwood
,
N.
,
Ffytche
,
D.
,
Simmons
,
A.
,
Bentall
,
R.
,
Murray
,
R.
, &
Howard
,
R.
(
2004
).
The cerebellum and decision making under uncertainty
.
Cognitive Brain Research
,
20
,
46
53
.
Blakemore
,
S. J.
,
Frith
,
C. D.
, &
Wolpert
,
D. M.
(
2001
).
The cerebellum is involved in predicting the sensory consequences of action
.
NeuroReport
,
12
,
1879
1884
.
Braine
,
M. D.
(
1978
).
On the relation between the natural logic of reasoning and standard logic
.
Psychological Review
,
85
,
1
21
.
Braine
,
M. D.
, &
O'Brien
,
D. P.
(
1991
).
A theory of if: A lexical entry, reasoning program, and pragmatic principles
.
Psychological Review
,
98
,
182
203
.
*
Brunetti
,
M.
,
Perrucci
,
M. G.
,
Di Naccio
,
M. R.
,
Ferretti
,
A.
,
Del Gratta
,
C.
,
Casadio
,
C.
, et al
(
2014
).
Framing deductive reasoning with emotional content: An fMRI study
.
Brain and Cognition
,
87
,
153
160
.
Bunge
,
S. A.
,
Kahn
,
I.
,
Wallis
,
J. D.
,
Miller
,
E. K.
, &
Wagner
,
A. D.
(
2003
).
Neural circuits subserving the retrieval and maintenance of abstract rules
.
Journal of Neurophysiology
,
90
,
3419
3428
.
*
Canessa
,
N.
,
Gorini
,
A.
,
Cappa
,
S. F.
,
Piattelli-Palmarini
,
M.
,
Danna
,
M.
,
Fazio
,
F.
, et al
(
2005
).
The effect of social content on deductive reasoning: An fMRI study
.
Human Brain Mapping
,
26
,
30
43
.
*
Canessa
,
N.
,
Pantaleo
,
G.
,
Crespi
,
C.
,
Gorini
,
A.
, &
Cappa
,
S. F.
(
2014
).
The impact of egocentric vs. allocentric agency attributions on the neural bases of reasoning about social rules
.
Brain Research
,
1581
,
40
50
.
Christoff
,
K.
,
Prabhakaran
,
V.
,
Dorfman
,
J.
,
Zhao
,
Z.
,
Kroger
,
J. K.
,
Holyoak
,
K. J.
, et al
(
2001
).
Rostrolateral prefrontal cortex involvement in relational integration during reasoning
.
Neuroimage
,
14
,
1136
1149
.
*
Coetzee
,
J. P.
, &
Monti
,
M. M.
(
2018
).
At the core of reasoning: Dissociating deductive and non-deductive load
.
Human Brain Mapping
,
39
,
1850
1861
.
Collins
,
D. L.
,
Neelin
,
P.
,
Peters
,
T. M.
, &
Evans
,
A. C.
(
1994
).
Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space
.
Journal of Computer Assisted Tomography
,
18
,
192
205
.
Cona
,
G.
, &
Scarpazza
,
C.
(
2019
).
Where is the “where” in the brain? A meta-analysis of neuroimaging studies on spatial cognition
.
Human Brain Mapping
,
40
,
1867
1886
.
Dennett
,
D. C.
(
1998
).
Reflections on language and mind
. In
P.
Carruthers
&
J.
Boucher
(Eds.),
Language and thought: Interdisciplinary themes
(pp.
284
294
).
Cambridge, United Kingdom
:
Cambridge University Press
.
De Pisapia
,
N.
, &
Braver
,
T. S.
(
2008
).
Preparation for integration: The role of anterior prefrontal cortex in working memory
.
NeuroReport
,
19
,
15
19
.
Eickhoff
,
S. B.
,
Bzdok
,
D.
,
Laird
,
A. R.
,
Kurth
,
F.
, &
Fox
,
P. T.
(
2012
).
Activation likelihood estimation meta-analysis revisited
.
Neuroimage
,
59
,
2349
2361
.
Eickhoff
,
S. B.
,
Bzdok
,
D.
,
Laird
,
A. R.
,
Roski
,
C.
,
Caspers
,
S.
,
Zilles
,
K.
, et al
(
2011
).
Co-activation patterns distinguish cortical modules, their connectivity and functional differentiation
.
Neuroimage
,
57
,
938
949
.
Eickhoff
,
S. B.
,
Laird
,
A. R.
,
Grefkes
,
C.
,
Wang
,
L. E.
,
Zilles
,
K.
, &
Fox
,
P. T.
(
2009
).
Coordinate-based activation likelihood estimation meta-analysis of neuroimaging data: A random-effects approach based on empirical estimates of spatial uncertainty
.
Human Brain Mapping
,
30
,
2907
2926
.
Eickhoff
,
S. B.
,
Nichols
,
T. E.
,
Laird
,
A. R.
,
Hoffstaedter
,
F.
,
Amunts
,
K.
,
Fox
,
P. T.
, et al
(
2016
).
Behavior, sensitivity, and power of activation likelihood estimation characterized by massive empirical simulation
.
Neuroimage
,
137
,
70
85
.
Embick
,
D.
,
Marantz
,
A.
,
Miyashita
,
Y.
,
O'Neil
,
W.
, &
Sakai
,
K. L.
(
2000
).
A syntactic specialization for Broca's area
.
Proceedings of the National Academy of Sciences, U.S.A.
,
97
,
6150
6154
.
Ermer
,
E.
,
Guerin
,
S. A.
,
Cosmides
,
L.
,
Tooby
,
J.
, &
Miller
,
M. B.
(
2006
).
Theory of mind broad and narrow: Reasoning about social exchange engages ToM areas, precautionary reasoning does not
.
Social Neuroscience
,
1
,
196
219
.
Evans
,
J. S. B. T.
,
Barston
,
J. L.
, &
Pollard
,
P.
(
1983
).
On the conflict between logic and belief in syllogistic reasoning
.
Memory & Cognition
,
11
,
295
306
.
Evans
,
A. C.
,
Collins
,
D. L.
,
Mills
,
S. R.
,
Brown
,
E. D.
,
Kelly
,
R. L.
, &
Peters
,
T. M.
(
1993
).
3D statistical neuroanatomical models from 305 MRI volumes
. In
L. A.
Klausner
(Ed.),
Nuclear Science Symposium and Medical Imaging Conference
(pp.
1813
1817
).
Piscataway, NJ
:
IEEE Service Centre
.
Evans
,
J. S. B. T.
, &
Over
,
D. E.
(
1996
).
Rationality and reasoning
.
Hove, United Kingdom
:
Psychology Press
.
Favrel
,
J.
, &
Barrouillet
,
P.
(
2000
).
On the relation between representations constructed from text comprehension and transitive inference production
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
26
,
187
203
.
Fiddick
,
L.
,
Spampinato
,
M. V.
, &
Grafman
,
J.
(
2005
).
Social contracts and precautions activate different neurological systems: An fMRI investigation of deontic reasoning
.
Neuroimage
,
28
,
778
786
.
Fletcher
,
P. C.
, &
Henson
,
R. N. A.
(
2001
).
Frontal lobes and human memory: Insights from functional neuroimaging
.
Brain
,
124
,
849
881
.
Fodor
,
J. A.
(
1983
).
The modularity of mind: An essay on faculty psychology
.
Cambridge, MA
:
MIT Press
.
Friederici
,
A. D.
(
2002
).
Towards a neural basis of auditory sentence processing
.
Trends in Cognitive Sciences
,
6
,
78
84
.
Friederici
,
A. D.
(
2011
).
The brain basis of language processing: From structure to function
.
Physiological Reviews
,
91
,
1357
1392
.
Friederici
,
A. D.
,
Fiebach
,
C. J.
,
Schlesewsky
,
M.
,
Bornkessel
,
I. D.
, &
von Cramon
,
D. Y.
(
2006
).
Processing linguistic complexity and grammaticality in the left frontal cortex
.
Cerebral Cortex
,
16
,
1709
1717
.
Friederici
,
A. D.
, &
Kotz
,
S. A.
(
2003
).
The brain basis of syntactic processes: Functional imaging and lesion studies
.
Neuroimage
,
20(Suppl. 1)
,
S8
S17
.
Friederici
,
A. D.
,
Kotz
,
S. A.
,
Werheid
,
K.
,
Hein
,
G.
, &
von Cramon
,
D. Y.
(
2003
).
Syntactic comprehension in Parkinson's disease: Investigating early automatic and late integrational processes using event-related brain potentials
.
Neuropsychology
,
17
,
133
142
.
Goel
,
V.
(
2003
).
Evidence for dual neural pathways for syllogistic reasoning
.
Psychologica
,
32
,
301
309
.
Goel
,
V.
(
2007
).
Anatomy of deductive reasoning
.
Trends in Cognitive Sciences
,
11
,
435
441
.
*
Goel
,
V.
,
Buchel
,
C.
,
Frith
,
C.
, &
Dolan
,
R. J.
(
2000
).
Dissociation of mechanisms underlying syllogistic reasoning
.
Neuroimage
,
12
,
504
514
.
*
Goel
,
V.
, &
Dolan
,
R. J.
(
2003
).
Explaining modulation of reasoning by belief
.
Cognition
,
87
,
B11
B22
.
*
Goel
,
V.
, &
Dolan
,
R. J.
(
2004
).
Differential involvement of left prefrontal cortex in inductive and deductive reasoning
.
Cognition
,
93
,
B109
B121
.
*
Goel
,
V.
,
Gold
,
B.
,
Kapur
,
S.
, &
Houle
,
S.
(
1997
).
The seats of reason? An imaging study of deductive and inductive reasoning
.
NeuroReport
,
8
,
1305
1310
.
*
Goel
,
V.
,
Gold
,
B.
,
Kapur
,
S.
, &
Houle
,
S.
(
1998
).
Neuroanatomical correlates of human reasoning
.
Journal of Cognitive Neuroscience
,
10
,
293
302
.
Goel
,
V.
,
Makale
,
M.
, &
Grafman
,
J.
(
2004
).
The hippocampal system mediates logical reasoning about familiar spatial environments
.
Journal of Cognitive Neuroscience
,
16
,
654
664
.
Goertzel
,
B.
(
2012
).
Perception processing for general intelligence: Bridging the symbolic/subsymbolic gap
. In
J.
Bach
,
B.
Goertzel
, &
M.
Iklé
(Eds.),
Proceedings of the 5th International Conference on Artificial General Intelligence
(pp.
79
88
).
Berlin, Heidelberg
:
Springer
.
Grodzinsky
,
Y.
, &
Friederici
,
A. D.
(
2006
).
Neuroimaging of syntax and syntactic processing
.
Current Opinion in Neurobiology
,
16
,
240
246
.
Grodzinsky
,
Y.
, &
Santi
,
A.
(
2008
).
The battle for Broca's region
.
Trends in Cognitive Sciences
,
12
,
474
480
.
Hagoort
,
P.
(
2005
).
On Broca, brain, and binding: A new framework
.
Trends in Cognitive Sciences
,
9
,
416
423
.
Hagoort
,
P.
,
Hald
,
L.
,
Bastiaansen
,
M.
, &
Petersson
,
K. M.
(
2004
).
Integration of word meaning and world knowledge in language comprehension
.
Science
,
304
,
438
441
.
*
Hearne
,
L.
,
Cocchi
,
L.
,
Zalesky
,
A.
, &
Mattingley
,
J. B.
(
2015
).
Interactions between default mode and control networks as a function of increasing cognitive reasoning complexity
.
Human Brain Mapping
,
36
,
2719
2731
.
Heit
,
E.
(
2015
).
Brain imaging, forward inference, and theories of reasoning
.
Frontiers in Human Neuroscience
,
8
,
1056
.
Henle
,
M.
(
1962
).
On the relation between logic and thinking
.
Psychological Review
,
69
,
366
378
.
Henson
,
R.
(
2006
).
Forward inference using functional neuroimaging: Dissociations versus associations
.
Trends in Cognitive Sciences
,
10
,
64
69
.
Hobeika
,
L.
,
Diard-Detoeuf
,
C.
,
Garcin
,
B.
,
Levy
,
R.
, &
Volle
,
E.
(
2016
).
General and specialized brain correlates for analogical reasoning: A meta-analysis of functional imaging studies
.
Human Brain Mapping
,
37
,
1953
1969
.
Hochstadt
,
J.
,
Nakano
,
H.
,
Lieberman
,
P.
, &
Friedman
,
J.
(
2006
).
The roles of sequencing and verbal working memory in sentence comprehension deficits in Parkinson's disease
.
Brain and Language
,
97
,
243
257
.
Houdé
,
O.
,
Zago
,
L.
,
Mellet
,
E.
,
Moutier
,
S.
,
Pineau
,
A.
,
Mazoyer
,
B.
, et al
(
2000
).
Shifting from the perceptual brain to the logical brain: The neural impact of cognitive inhibition training
.
Journal of Cognitive Neuroscience
,
12
,
721
728
.
Ito
,
M.
(
1993
).
Movement and thought: Identical control mechanisms by the cerebellum
.
Trends in Neurosciences
,
16
,
448
450
.
*
Jia
,
X.
,
Lu
,
S.
,
Zhong
,
N.
,
Yao
,
Y.
,
Li
,
K.
, &
Yang
,
Y.
(
2009
).
Common and distinct neural substrates of forward-chaining and backward-chaining syllogistic reasoning
. In
International Conference on Complex Medical Engineering, ICME 2009
(pp.
1
6
).
Tempe, AZ
:
IEEE
.
Johnson-Laird
,
P. N.
(
1980
).
Mental models in cognitive science
.
Cognitive Science
,
4
,
71
115
.
Johnson-Laird
,
P. N.
(
1983
).
Mental models: Towards a cognitive science of language, inference, and consciousness
.
Cambridge, MA
:
Harvard University Press
.
Johnson-Laird
,
P. N.
(
2006
).
How we reason
.
New York
:
Oxford University Press
.
Keren
,
G.
(
2013
).
A tale of two systems: A scientific advance or a theoretical stone soup? Commentary on Evans & Stanovich (2013)
.
Perspectives on Psychological Science
,
8
,
257
262
.
Khemlani
,
S.
, &
Johnson-Laird
,
P. N.
(
2012
).
Theories of the syllogism: A meta-analysis
.
Psychological Bulletin
,
138
,
427
457
.
Knauff
,
M.
,
Fangmeier
,
T.
,
Ruff
,
C. C.
, &
Johnson-Laird
,
P. N.
(
2003
).
Reasoning, models, and images: Behavioral measures and cortical activity
.
Journal of Cognitive Neuroscience
,
15
,
559
573
.
Knauff
,
M.
,
Mulack
,
T.
,
Kassubek
,
J.
,
Salih
,
H. R.
, &
Greenlee
,
M. W.
(
2002
).
Spatial imagery in deductive reasoning: A functional MRI study
.
Cognitive Brain Research
,
13
,
203
212
.
Koechlin
,
E.
,
Corrado
,
G.
,
Pietrini
,
P.
, &
Grafman
,
J.
(
2000
).
Dissociating the role of the medial and lateral anterior prefrontal cortex in human planning
.
Proceedings of the National Academy of Sciences, U.S.A.
,
97
,
7651
7656
.
Kotz
,
S. A.
,
Frisch
,
S.
,
von Cramon
,
D. Y.
, &
Friederici
,
A. D.
(
2003
).
Syntactic language processing: ERP lesion data on the role of the basal ganglia
.
Journal of the International Neuropsychological Society
,
9
,
1053
1060
.
*
Kroger
,
J. K.
,
Nystrom
,
L. E.
,
Cohen
,
J. D.
, &
Johnson-Laird
,
P. N.
(
2008
).
Distinct neural substrates for deductive and mathematical processing
.
Brain Research
,
1243
,
86
103
.
Kroger
,
J. K.
,
Sabb
,
F. W.
,
Fales
,
C. L.
,
Bookheimer
,
S. Y.
,
Cohen
,
M. S.
, &
Holyoak
,
K. J.
(
2002
).
Recruitment of anterior dorsolateral prefrontal cortex in human reasoning: A parametric study of relational complexity
.
Cerebral Cortex
,
12
,
477
485
.
Laird
,
A. R.
,
Robinson
,
J. L.
,
McMillan
,
K. M.
,
Tordesillas-Gutiérrez
,
D.
,
Moran
,
S. T.
,
Gonzales
,
S. M.
, et al
(
2010
).
Comparison of the disparity between Talairach and MNI coordinates in functional neuroimaging data: Validation of the Lancaster transform
.
Neuroimage
,
51
,
677
683
.
Lemaire
,
J.-J.
,
Golby
,
A.
,
Wells
,
W. M.
, III
,
Pujol
,
S.
,
Tie
,
Y.
,
Rigolo
,
L.
, et al
(
2013
).
Extended Broca's area in the functional connectome of language in adults: Combined cortical and subcortical single-subject analysis using fMRI and DTI tractography
.
Brain Topography
,
26
,
428
441
.
*
Liu
,
J. M.
,
Zhang
,
M.
,
Jou
,
J.
,
Wu
,
X.
,
Li
,
W.
, &
Qiu
,
J.
(
2012
).
Neural bases of falsification in conditional proposition testing: Evidence from an fMRI study
.
International Journal of Psychophysiology
,
85
,
249
256
.
Marr
,
D.
(
1982
).
Vision: A computational investigation into the human representation and processing of visual information
.
San Francisco
:
W. H. Freeman
.
Mellet
,
E.
,
Petit
,
L.
,
Mazoyer
,
B.
,
Denis
,
M.
, &
Tzourio
,
N.
(
1998
).
Reopening the mental imagery debate: Lessons from functional anatomy
.
Neuroimage
,
8
,
129
139
.
Meyer
,
M.
,
Friederici
,
A. D.
, &
von Cramon
,
D. Y.
(
2000
).
Neurocognition of auditory sentence comprehension: Event related fMRI reveals sensitivity to syntactic violations and task demands
.
Cognitive Brain Research
,
9
,
19
33
.
Monti
,
M. M.
, &
Osherson
,
D. N.
(
2012
).
Logic, language and the brain
.
Brain Research
,
1428
,
33
42
.
*
Monti
,
M. M.
,
Osherson
,
D. N.
,
Martinez
,
M. J.
, &
Parsons
,
L. M.
(
2007
).
Functional neuroanatomy of deductive inference: A language-independent distributed network
.
Neuroimage
,
37
,
1005
1016
.
*
Monti
,
M. M.
,
Parsons
,
L. M.
, &
Osherson
,
D. N.
(
2009
).
The boundaries of language and thought in deductive inference
.
Proceedings of the National Academy of Sciences, U.S.A.
,
106
,
12554
12559
.
Moro
,
A.
,
Tettamanti
,
M.
,
Perani
,
D.
,
Donati
,
C.
,
Cappa
,
S. F.
, &
Fazio
,
F.
(
2001
).
Syntax and the brain: Disentangling grammar by selective anomalies
.
Neuroimage
,
13
,
110
118
.
Newman
,
S. D.
,
Just
,
M. A.
,
Keller
,
T. A.
,
Roth
,
J.
, &
Carpenter
,
P. A.
(
2003
).
Differential effects of syntactic and semantic processing on the subregions of Broca's area
.
Cognitive Brain Research
,
16
,
297
307
.
Ni
,
W.
,
Constable
,
R. T.
,
Mencl
,
W. E.
,
Pugh
,
K. R.
,
Fulbright
,
R. K.
,
Shaywitz
,
S. E.
, et al
(
2000
).
An event-related neuroimaging study distinguishing form and content in sentence processing
.
Journal of Cognitive Neuroscience
,
12
,
120
133
.
*
Noveck
,
I. A.
,
Goel
,
V.
, &
Smith
,
K. W.
(
2004
).
The neural basis of conditional reasoning with arbitrary content
.
Cortex
,
40
,
613
622
.
Oaksford
,
M.
(
2015
).
Imaging deductive reasoning and the new paradigm
.
Frontiers in Human Neuroscience
,
9
,
101
.
Oaksford
,
M.
, &
Chater
,
N.
(
2001
).
The probabilistic approach to human reasoning
.
Trends in Cognitive Sciences
,
5
,
349
357
.
Oaksford
,
M.
, &
Chater
,
N.
(
2007
).
Bayesian rationality: The probabilistic approach to human reasoning
.
Oxford, United Kingdom
:
Oxford University Press
.
Oberauer
,
K.
(
2006
).
Reasoning with conditionals: A test of formal models of four theories
.
Cognitive Psychology
,
53
,
238
283
.
*
Osherson
,
D. N.
,
Perani
,
D.
,
Cappa
,
S.
,
Schnur
,
T.
,
Grassi
,
F.
, &
Fazio
,
F.
(
1998
).
Distinct brain loci in deductive versus probabilistic reasoning
.
Neuropsychologia
,
36
,
369
376
.
*
Parsons
,
L. M.
, &
Osherson
,
D. N.
(
2001
).
New evidence for distinct right and left brain systems for deductive versus probabilistic reasoning
.
Cerebral Cortex
,
11
,
954
965
.
Petrides
,
M.
(
1995
).
Impairments on nonspatial self-ordered and externally ordered working memory tasks after lesions of the mid-dorsal part of the lateral frontal cortex in the monkey
.
Journal of Neuroscience
,
15
,
359
375
.
Pickett
,
E. R.
,
Kuniholm
,
E.
,
Protopapas
,
A.
,
Friedman
,
J.
, &
Lieberman
,
P.
(
1998
).
Selective speech motor, syntax and cognitive deficits associated with bilateral damage to the putamen and the head of the caudate nucleus: A case study
.
Neuropsychologia
,
36
,
173
188
.
*
Porcaro
,
C.
,
Medaglia
,
M. T.
,
Thai
,
N. J.
,
Seri
,
S.
,
Rotshtein
,
P.
, &
Tecchio
,
F.
(
2014
).
Contradictory reasoning network: An EEG and fMRI study
.
PLoS One
,
9
,
e92835
.
Posner
,
M. I.
, &
Dehaene
,
S.
(
1994
).
Attentional networks
.
Trends in Neurosciences
,
17
,
75
79
.
Prado
,
J.
(
2018
).
The relationship between deductive reasoning and the syntax of language in Broca's area: A review of the neuroimaging literature
.
L'année Psychologique
,
118
,
289
315
.
Prado
,
J.
,
Chadha
,
A.
, &
Booth
,
J. R.
(
2011
).
The brain network for deductive reasoning: A quantitative meta-analysis of 28 neuroimaging studies
.
Journal of Cognitive Neuroscience
,
23
,
3483
3497
.
*
Prado
,
J.
, &
Noveck
,
I. A.
(
2007
).
Overcoming perceptual features in logical reasoning: A parametric functional magnetic resonance imaging study
.
Journal of Cognitive Neuroscience
,
19
,
642
657
.
*
Prado
,
J.
,
Van Der Henst
,
J.-B.
, &
Noveck
,
I. A.
(
2010
).
Recomposing a fragmented literature: How conditional and relational arguments engage different neural systems for deductive reasoning
.
Neuroimage
,
51
,
1213
1221
.
Rao
,
S. M.
,
Bobholz
,
J. A.
,
Hammeke
,
T. A.
,
Rosen
,
A. C.
,
Woodley
,
S. J.
,
Cunningham
,
J. M.
, et al
(
1997
).
Functional MRI evidence for subcortical participation in conceptual reasoning skills
.
NeuroReport
,
8
,
1987
1993
.
*
Reverberi
,
C.
,
Bonatti
,
L. L.
,
Frackowiak
,
R. S. J.
,
Paulesu
,
E.
,
Cherubini
,
P.
, &
Macaluso
,
E.
(
2012
).
Large scale brain activations predict reasoning profiles
.
Neuroimage
,
59
,
1752
1764
.
*
Reverberi
,
C.
,
Cherubini
,
P.
,
Frackowiak
,
R. S. J.
,
Caltagirone
,
C.
,
Paulesu
,
E.
, &
Macaluso
,
E.
(
2010
).
Conditional and syllogistic deductive tasks dissociate functionally during premise integration
.
Human Brain Mapping
,
31
,
1430
1445
.
*
Reverberi
,
C.
,
Cherubini
,
P.
,
Rapisarda
,
A.
,
Rigamonti
,
E.
,
Caltagirone
,
C.
,
Frackowiak
,
R. S. J.
, et al
(
2007
).
Neural basis of generation of conclusions in elementary deduction
.
Neuroimage
,
38
,
752
762
.
Reverberi
,
C.
,
Shallice
,
T.
,
D'Agostini
,
S.
,
Skrap
,
M.
, &
Bonatti
,
L. L.
(
2009
).
Cortical bases of elementary deductive reasoning: Inference, memory, and metadeduction
.
Neuropsychologia
,
47
,
1107
1116
.
Rips
,
L. J.
(
1994
).
The psychology of proof: Deductive reasoning in human thinking
.
Cambridge, MA
:
MIT Press
.
Rips
,
L. J.
(
2001
).
Two kinds of reasoning
.
Psychological Science
,
12
,
129
134
.
*
Rodriguez-Moreno
,
D.
, &
Hirsch
,
J.
(
2009
).
The dynamics of deductive reasoning: An fMRI investigation
.
Neuropsychologia
,
47
,
949
961
.
Schmahmann
,
J. D.
(
1991
).
An emerging concept: The cerebellar contribution to higher function
.
Archives of Neurology
,
48
,
1178
1187
.
Schmahmann
,
J. D.
(
1996
).
From movement to thought: Anatomic substrates of the cerebellar contribution to cognitive processing
.
Human Brain Mapping
,
4
,
174
198
.
*
Smith
,
K. W.
,
Balkwill
,
L.-L.
,
Vartanian
,
O.
, &
Goel
,
V.
(
2015
).
Syllogisms delivered in an angry voice lead to improved performance and engagement of a different neural system compared to neutral voice
.
Frontiers in Human Neuroscience
,
9
,
273
.
*
Smith
,
K. W.
,
Vartanian
,
O.
, &
Goel
,
V.
(
2014
).
Dissociable neural systems underwrite logical reasoning in the context of induced emotions with positive and negative valence
.
Frontiers in Human Neuroscience
,
8
,
736
.
Stanovich
,
K. E.
, &
West
,
R. F.
(
2000
).
Individual differences in reasoning: Implications for the rationality debate?
Behavioral and Brain Sciences
,
23
,
645
665
.
Talairach
,
J.
, &
Tournoux
,
P.
(
1988
).
Co-planar stereotaxic atlas of the human brain. 3-dimensional proportional system: An approach to cerebral imaging
.
New York
:
Thieme
.
Tettamanti
,
M.
,
Moro
,
A.
,
Messa
,
C.
,
Moresco
,
R. M.
,
Rizzo
,
G.
,
Carpinelli
,
A.
, et al
(
2005
).
Basal ganglia and language: Phonology modulates dopaminergic release
.
NeuroReport
,
16
,
397
401
.
Turkeltaub
,
P. E.
,
Eickhoff
,
S. B.
,
Laird
,
A. R.
,
Fox
,
M.
,
Wiener
,
M.
, &
Fox
,
P.
(
2012
).
Minimizing within-experiment and within-group effects in activation likelihood estimation meta-analyses
.
Human Brain Mapping
,
33
,
1
13
.
Ullman
,
M. T.
(
2006
).
Is Broca's area part of a basal ganglia thalamocortical circuit?
Cortex
,
42
,
480
485
.
van den Heuvel
,
O. A.
,
Groenewegen
,
H. J.
,
Barkhof
,
F.
,
Lazeron
,
R. H. C.
,
van Dyck
,
R.
, &
Veltman
,
D. J.
(
2003
).
Frontostriatal system in planning complexity: A parametric functional magnetic resonance version of Tower of London task
.
Neuroimage
,
18
,
367
374
.
*
Van Hoeck
,
N.
,
Begtas
,
E.
,
Steen
,
J.
,
Kestemont
,
J.
,
Vandekerckhove
,
M.
, &
Van Overwalle
,
F.
(
2014
).
False belief and counterfactual reasoning in a social environment
.
Neuroimage
,
90
,
315
325
.
Volz
,
K. G.
,
Schubotz
,
R. I.
, &
von Cramon
,
D. Y.
(
2005
).
Variants of uncertainty in decision-making and their neural correlates
.
Brain Research Bulletin
,
67
,
403
412
.
Wason
,
P.
(
1966
).
Reasoning
. In
B. M.
Foss
(Ed.),
New horizons in psychology
(
Vol. 1
, pp.
135
151
).
Harmondsworth, United Kingdom
:
Penguin
.
Wertheim
,
J.
, &
Ragni
,
M.
(
2018
).
The neural correlates of relational reasoning: A meta-analysis of 47 functional magnetic resonance studies
.
Journal of Cognitive Neuroscience
,
30
,
1734
1748
.