Raven's Progressive Matrices Research Articles

Measuring Raven’s Progressive Matrices Combining Eye-Tracking Technology and Machine Learning (ML) Models

Extended testing time in Raven’s Progressive Matrices (RPM) can lead to increased fatigue and reduced motivation, which may impair cognitive task performance. This study explores the application of artificial intelligence (AI) in RPM by combining eye-tracking technology with machine learning (ML) models, aiming to explore new methods for improving the efficiency of RPM testing and to identify the key metrics involved. Using eye-tracking metrics as features, ten ML models were trained, with the XGBoost model demonstrating superior performance. Notably, we further refined the period of interest and reduced the number of metrics, achieving strong performance, with accuracy, precision, and recall all above 0.8, using only 60% of the response time and nine eye-tracking metrics. This study also examines the role of several key metrics in RPM and offers valuable insights for future research.

Two-stage Rule-induction visual reasoning on RPMs with an application to video prediction

Raven’s Progressive Matrices (RPMs) are frequently used in evaluating human’s visual reasoning ability. Researchers have made considerable efforts in developing systems to automatically solve the RPM problem, often through a black-box end-to-end convolutional neural network for both visual recognition and logical reasoning tasks. Based on the intrinsic natures of RPM problem, we propose a Two-stage Rule-Induction Visual Reasoner (TRIVR), which consists of a perception module and a reasoning module, to tackle the challenges of real-world visual recognition and subsequent logical reasoning tasks, respectively. For the reasoning module, we further propose a “2+1” formulation that models human’s thinking in solving RPMs and significantly reduces the model complexity. It derives a reasoning rule from each RPM sample, which is not feasible for existing methods. As a result, the proposed reasoning module is capable of yielding a set of reasoning rules modeling human in solving the RPM problems. To validate the proposed method on real-world applications, an RPM-like Video Prediction (RVP) dataset is constructed, where visual reasoning is conducted on RPMs constructed using real-world video frames. Experimental results on various RPM-like datasets demonstrate that the proposed TRIVR achieves a significant and consistent performance gain compared with state-of-the-art models.

Mental imagery in education: What impact on the relationships with visuospatial processing and school performance in junior high school students?

This study examined the relationship between mental imagery, visuospatial processing, and academic performance among junior high school students. The influence of these variables on learning outcomes in Italian, mathematics, art, and music was also investigated. Third-year students were tested using the Mental Imagery Test and Raven's Progressive Matrices. Gender differences were explored, revealing no significant differences in mental imagery and visuospatial processing. Higher mental imagery correlated with better academic performance, particularly in Italian and mathematics. Visuospatial processing mediated the relationship between mental imagery and academic performance in Italian and mathematics, suggesting its significant role in these subjects. However, this mediation was not significant for music and art, suggesting the involvement of alternative cognitive mechanisms. These findings highlight the importance of mental imagery and visuospatial processing in academic success, particularly in abstract subjects, and advocate for educational strategies targeting these cognitive domains to enhance not only student performance but also creativity and thinking skills.

Influence of problem-solving ability and personality variables on the improvement and creativity of tactical decisions in basketball.

Basketball players are increasingly required not only to read the tactical actions that take place in a game but also to make correct tactical decisions. This includes greater creativity in the type of solutions they must provide when faced with various tactical situations. To acquire these skills, young players need training in which personality and general intelligence variables significantly influence the learning process. A Perceptual Tactical Training (PTT) program was implemented, correlated with the Eysenck Personality Questionnaire and the Raven's Progressive Matrices test, to observe improvements in tactical perception and decision-making among young basketball players. The PTT produced significant improvements in tactical solutions and tactical perception for all players. Players who scored high in general intelligence found it easier to perceive tactical situations but struggled to generate original tactical solutions. Conversely, players with high psychoticism and high general intelligence were more capable of creating innovative tactical solutions but did not excel in perceiving tactical game situations. Therefore, it is crucial to be aware of the individual differences in intelligence and personality traits among young players to enhance their tactical decision-making and creative abilities.

Beyond traditional boundaries: Exploring the jigsaw method to strengthen academic achievement in biology

Cooperative teaching methods have emerged as a significant paradigm shift from traditional teaching methods, highlighting notable benefits that have a profound influence on academic achievement. Guided by previous studies on cooperative teaching methods, we examined the effectiveness of teaching methods (jigsaw vs. lecture-cum-demonstration methods) on the academic achievement of biology among class 11 students. The sample comprised 68 students in class 11. In a school setting, 35 participants were randomly assigned to the experimental group, and 33 were assigned to the control group. The study utilized a quasiexperimental research design to assess the effectiveness of different teaching methods. Specifically, a nonequivalent control group design was employed, comprising one experimental group and one control group. To ensure comparability between the groups in terms of academic performance, Raven's Progressive Matrices Intelligence test score was used as a covariate. The results revealed that class 11 biology students in the experimental group had notably higher test scores than did those in the control group. Additionally, the study revealed that gender did not significantly influence academic achievement in biology for class 11 students when intelligence was used as a covariate. The academic achievement of class 11 biology students appears to be independent of the interaction between gender and method of teaching. We explicitly demonstrate that the jigsaw technique can enhance the biology achievement of senior secondary school students and can serve as an effective alternative to traditional experimental teaching methods in India.

The Impact of Exposure to Iodine and Fluorine in Drinking Water on Thyroid Health and Intelligence in School-Age Children: A Cross-Sectional Investigation.

Iodine and fluorine, as halogen elements, are often coexisting in water environments, with nearly 200 million people suffering from fluorosis globally, and, in 11 countries and territories, adolescents have iodine intakes higher than that required for the prevention of iodine deficiency disorders. It has been suggested that excess iodine and/or fluorine can affect thyroid health and intellectual development, especially in children, but their combined effect has been less studied in this population. This study investigated 399 school-age children in Tianjin, China, collected drinking water samples from areas where the school-age children lived, and grouped the respondents according to iodine and fluorine levels. Thyroid health was measured using thyroid hormone levels, thyroid volume, and the presence of thyroid nodules; intelligence quotient (IQ) was assessed using the Raven's Progressive Matrices (CRT) test; and monoamine neurotransmitter levels were used to explore the potential relationship between thyroid health and intelligence. Multiple linear regression and restricted cubic spline (RCS) analyses showed that iodine and fluorine were positively correlated with thyroid volume and the incidence of thyroid nodules in school-age children, and negatively correlated with IQ; similar results were obtained in the secondary subgroups based on urinary iodine and urinary fluoride levels. Interaction analyses revealed a synergistic effect of iodine and fluorine. A pathway analysis showed that iodine and fluorine were negatively associated with the secretion of free triiodothyronine (FT3) and free tetraiodothyronine (FT4), which in turn were negatively associated with the secretion of thyroid-stimulating hormone (TSH). Iodine and fluorine may affect IQ in school-aged children through the above pathways that affect thyroid hormone secretion; of these, FT3 and TSH were negatively correlated with IQ, whereas FT4 was positively correlated with IQ. The relationship between thyroid hormones and monoamine neurotransmitters may involve the hypothalamic-pituitary-thyroid axis, with FT4 hormone concentrations positively correlating with dopamine (DA), norepinephrine (NE), and 5-hydroxytryptophan (5-HT) concentrations, and FT3 hormone concentrations positively correlating with DA concentrations. Monoamine neurotransmitters may play a mediating role in the effects of iodine and fluoride on intelligence in schoolchildren. However, this study has some limitations, as the data were derived from a cross-sectional study in Tianjin, China, and no attention was paid to the reciprocal effects of iodine and fluorine at different doses on thyroid health and intelligence in schoolchildren in other regions.

Fluid intelligence evaluation through raven's progressive matrices in adult patients with congenital hypothyroidism detected by neonatal screening

Fluid intelligence evaluation through raven's progressive matrices in adult patients with congenital hypothyroidism detected by neonatal screening

Cognition in Patients with Spinocerebellar Ataxia 1 (SCA1) and 2 (SCA2): A Neurophysiological and Neuropsychological Approach.

Background/Objectives: Cognitive impairment in spinocerebellar ataxia patients has been reported since the early-disease stage. We aimed to assess cognitive differences in SCA1 and SCA2 patients. Methods: We performed neuropsychological (NPS) and neurophysiological (auditory event-related potentials, aERPs) assessments in 16 SCA1 and 18 SCA2 consecutive patients. Furthermore, clinical information (age at onset, disease duration, motor disability) was collected. Results: NPS tests yielded scores in the normal range in both groups but with lower scores in the Frontal Assessment Battery (p < 0.05) and Visual Analogue Test for Anosognosia for motor impairment (p < 0.05) in SCA1, and the Trail Making Test (p < 0.01), Raven's progressive matrices (p < 0.01), Stroop (p < 0.05), and emotion attribution tests (p < 0.05) in SCA2. aERPs showed lower N100 amplitude (p < 0.01) and prolonged N200 latency (p < 0.01) in SCA1 compared with SCA2. Clinically, SCA2 had more severe motor disability than SCA1 in the Assessment and Rating of Ataxia Scale. Conclusions: SCA2 showed more significant difficulties in attentional, visuospatial, and emotional function, and greater motor impairment. In contrast, SCA1 showed less cognitive flexibility/phasic ability, probably affected by a more severe degree of dysarthria. The same group revealed less neural activity during nonconscious attentional processing (N100-N200 data), suggesting greater involvement of sensory pathways in discriminating auditory stimuli. NFS did not correlate with NPS findings, implying an independent relationship. However, the specific role of the cerebellum and cerebellar symptoms in NPS test results deserves more focus.

Integrating music, drama and visual arts in extracurricular programs: enhancing psychological development in early school-aged children

The "Creative Bridges" study examines the integration of music, drama, and visual arts into extracurricular programs, focusing on its impact on the psychological development of early school-aged children. It addresses the need for holistic educational approaches that cater to cognitive, emotional, and social growth. This mixed-methods research utilized psychodiagnostic tools such as Raven's Progressive Matrices, the Academic Performance Rating Scale (APRS), the Strengths and Difficulties Questionnaire (SDQ), the Social Skills Improvement System (SSIS), and the Peer Relationship Questionnaire (PRQ) to assess a cohort of 64 children aged 11-12, divided into experimental and control groups. The study spanned six months, featuring pre-program, mid-program, post-program, and follow-up assessments. The program demonstrated significant improvements in non-verbal abstract reasoning, academic performance in math and reading, and emotional well-being. It also indicated nuanced effects on social skills and peer relationships, with initial disruptions giving way to eventual enhancements in social interactions. The findings illuminate the cognitive, emotional, and social benefits of arts integration, underscoring the potential of such programs to enrich educational curricula. The discussion delves into the implications of these results for educational practices and policies, advocating for the inclusion of the arts as central to curriculum design. The "Creative Bridges" program highlights the indispensable role of music, drama, and visual arts integration in fostering comprehensive psychological development among children. The study advocates for a broader implementation of arts integration in education, emphasizing its capacity to enhance cognitive abilities, emotional resilience, and social competencies.

A Controlled Clinical Trial on the Effects of Aquatic Exercise on Cognitive Functions in Community-Dwelling Older Adults.

Several therapies have been developed to reduce cognitive decline associated with aging. Aquatic exercises, which are widely used to enhance functional capacity, may play a role in stimulating cognitive functions. This study investigated the effects of a 3-month aquatic exercise program on cognitive functions in community-dwelling older adults. In this prospective, single-blinded, controlled clinical trial, 31 participants were allocated to either the experimental (aquatic exercises) or control (no-exercise) group. The intervention program consisted of exercises conducted twice a week in a 1.2 m deep indoor pool. The main outcome measures were cognitive functions, assessed using Raven's Progressive Matrices test and the Wisconsin Card Sorting Test. A repeated-measures analysis of variance was used to assess the impact of the exercise program. The effect sizes (η2p) were reported when a level of significance was achieved (p < 0.05). Compared with the control group, the participants who underwent aquatic exercises showed positive outcomes in Raven's Progressive Matrices test (p = 0.046; η2p = 0.131) and the Wisconsin Card Sorting Test (p = 0.001, η2p = 0.589). Complementary analyses of the Wisconsin Card Sorting Test indicated that the benefits of the aquatic exercise were observed in terms of the number of trials (p = 0.001, η2p = 0.478), number of errors (p = 0.001, η2p = 0.458), and number of non-perseverative errors (p = 0.001, η2p = 0.302). The results indicate that a period of three months of aquatic exercise was beneficial for stimulating specific aspects of the cognitive function of community-dwelling older individuals. Aquatic exercise should be prescribed to this population.

Retracted: The Relationship Between Oral and Written Language in Narrative Production by Arabic-Speaking Children: Fundamental Skills and Influences.

https://doi.org/10.1044/2024_Nov2024ASHA. This study aims to investigate the relationship between oral and written language skills in narrative production among Arabic-speaking children, focusing on cognitive and linguistic abilities. It examines the differences in narrative parameters between oral and written narratives and explores the associations between these parameters and cognitive and linguistic skills. The research involved 237 sixth-grade Arabic-speaking students from low-socioeconomic status schools in Israel. Each participant was instructed to orally tell a narrative and to write another narrative based on two sets of six sequential pictures. Various narrative features were analyzed, including word count for length, type-token ratio (TTR) for lexical diversity, mean length of utterance (MLU) for morphosyntax, and number of episodes for macrostructure. Cognitive linguistic measures, including Raven's Progressive Matrices, reading comprehension (RC), and morphological awareness (MA) were also assessed. The study found significant differences between oral and written narratives regarding lexical diversity and macrostructure. Participants exhibited significantly higher TTR in written narratives compared to oral narratives, whereas the number of episodes was significantly higher in oral narratives than in written ones. However, no significant differences were observed in narrative length or MLU. Moreover, the study identified significant predictors for various aspects of written narratives, particularly MA and RC, which significantly predicted TTR, MLU, and macrostructure. Additionally, the inclusion of word count in oral narratives significantly enhanced the explained variance for narrative length and macrostructure in written language. The results highlight the importance of the oral-written interface in both micro- and macrostructure representations in both oral and written modalities. They suggest that cognitive and linguistic skills, such as MA and RC, play a crucial role in narrative production. The findings have implications for educational practices and literacy outcomes in the Arab world, enhancing the understanding of the challenges and strategies involved in written language production among Arabic-speaking children.

Exploring the Neuropsychological Correlates of Swallowing Disorders in People with Parkinson's Disease: a Cross-Sectional Study.

Around 80% of persons with Parkinson's disease (PD) present symptoms of dysphagia. Although cognitive impairment may contribute to dysphagia, few studies have investigated the association between the PD neuropsychological profile and objective measures of swallowing dysfunction. Since the swallowing function comprises involuntary but also voluntary actions, we hypothesize that specific measures of attention and executive functions can be underlined in PD-related dysphagia. Therefore, the aim of this study was to extensively investigate the correlation and the relationship between attentive and executive functions and safety/efficiency of pharyngeal phase of swallowing in people with PD. All participants received a fiberoptic endoscopic evaluation of swallowing and were evaluated using the Penetration Aspiration Scale (PAS); the Yale Pharyngeal Residue Severity Rating Scale (IT-YPRSRS), and the Functional Oral Intake Scale (FOIS-IT). Participants also underwent a neuropsychological assessment covering global cognitive status, attention, and frontal executive functions. Correlations and associations between neuropsychological measures and swallowing components were calculated. Twenty-one participants with PD (mean age 69.38 ± 6.58 years, mean disease duration 8.38 ± 5.31 years; mean MDS-UPDRS III 43.95 ± 24.18) completed all evaluations. The most significant correlations were found between attentive functions (i.e., Stroop Time), and executive functions (i.e., Raven's Progressive Matrices, Digit Backward and Semantic Fluency), and FOIS-IT, PAS, and IT-YPRSRS sinuses and valleculae. These associations were not influenced by disease duration. These results suggest that a dysfunction to attentional processes and/or to executive functions can contribute to penetration and the presence of pharyngeal residue in participants with middle-stage PD.

Advancing with age: Older adults excel in comprehension of novel metaphors.

Older adults may experience certain forms of cognitive decline, but some forms of semantic memory remain intact in older age. To address how metaphor comprehension changes with age and whether metaphor comprehension relies more heavily on analogical reasoning (supported by fluid intelligence) or on conceptual combination (supported by crystalized intelligence), we compared performance of younger and older adults. In two experiments, healthy older adults (54-88 years) scored lower on a measure of fluid intelligence (Ravens Progressive Matrices) but higher on a measure of crystalized intelligence (Mill Hill Vocabulary Test) relative to younger adults (18-34 years). Groups were equally successful in comprehending relatively easy metaphors (Study 1), but older adults showed a striking advantage over younger adults for novel literary metaphors (Study 2). Mixed-effects modeling showed that measures of fluid and crystalized intelligence each made separable contributions to metaphor comprehension for both groups, but older adults relied more on crystalized intelligence than did younger adults. These age-related dissociations clarify cognitive effects of aging and highlight the importance of crystalized intelligence for metaphor comprehension in both younger and older adults. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Associations of childhood, adolescence, and midlife cognitive function with DNA methylation age acceleration in midlife.

Prior studies showed increased age acceleration (AgeAccel) is associated with worse cognitive function among old adults. We examine the associations of childhood, adolescence and midlife cognition with AgeAccel based on DNA methylation (DNAm) in midlife. Data are from 359 participants who had cognition measured in childhood and adolescence in the Child Health and Development study, and had cognition, blood based DNAm measured during midlife in the Disparities study. Childhood cognition was measured by Raven's Progressive Matrices and Peabody Picture Vocabulary Test (PPVT). Adolescent cognition was measured only by PPVT. Midlife cognition included Wechsler Test of Adult Reading (WTAR), Verbal Fluency (VF), Digit Symbol (DS). AgeAccel measures including Horvath, Hannum, PhenoAge, GrimAge and DunedinPACE were calculated from DNAm. Linear regressions adjusted for potential confounders were utilized to examine the association between each cognitive measure in relation to each AgeAccel. There are no significant associations between childhood cognition and midlife AgeAccel. A 1-unit increase in adolescent PPVT, which measures crystalized intelligence, is associated with 0.048-year decrease of aging measured by GrimAge and this association is attenuated after adjustment for adult socioeconomic status. Midlife crystalized intelligence measure WTAR is negatively associated with PhenoAge and DunedinPACE, and midlife fluid intelligence measure (DS) is negatively associated with GrimAge, PhenoAge and DunedinPACE. AgeAccel is not associated with VF in midlife. In conclusion, our study showed the potential role of cognitive functions at younger ages in the process of biological aging. We also showed a potential relationship of both crystalized and fluid intelligence with aging acceleration.

Beyond the heart: Cognitive and verbal outcomes in Arab children with congenital heart diseases.

This study investigates how congenital heart diseases (CHD) characteristics and interventions affect cognitive and verbal skills in Arab children, while also uncovering previously unexplored connections between these skills and the quality of life (QoL) scores as perceived by both children and parents. A cross-sectional study was conducted in Jordan, involving 62 children with CHD aged 6-16. Data were collected through standardized intelligence tests (namely The Raven's Progressive Matrices Test and The Wechsler Intelligence Scale for Children) and QoL assessments. Sex, disease severity, cyanosis, CHD defect status, conducted operations, and types of interventions did not significantly influence cognitive scores. However, a significant difference was observed in Wechsler's scores between cyanotic and non-cyanotic children (p < .01) and between severe and moderate cases (p = .01). Further, a significant positive correlation was identified between Wechsler's Scores and QoL reported by parents (r = 0.33, p < .01). This correlation was particularly pronounced in the social and school functioning dimensions of QoL. This study highlights the need for personalized care approaches for children with CHDs based on their individual characteristics. While cognitive abilities did not directly correlate with children's QoL reports, a significant positive correlation between verbal skills and QoL reported by parents underscores the importance of effective communication in assessing a child's overall well-being. Future research should further examine the cognitive development in this population, employing neurocognitive investigations and longitudinal studies to gain a deeper understanding of their cognitive profiles and trajectories.

Neuromodulatory effects of parietal high-definition transcranial direct-current stimulation on network-level activity serving fluid intelligence.

Fluid intelligence (Gf) involves rational thinking skills and requires the integration of information from different cortical regions to resolve novel complex problems. The effects of non-invasive brain stimulation on Gf have been studied in attempts to improve Gf, but such studies are rare and the few existing have reached conflicting conclusions. The parieto-frontal integration theory of intelligence (P-FIT) postulates that the parietal and frontal lobes play a critical role in Gf. To investigate the suggested role of parietal cortices, we applied high-definition transcranial direct current stimulation (HD-tDCS) to the left and right parietal cortices of 39 healthy adults (age 19-33years) for 20min in three separate sessions (left active, right active and sham). After completing the stimulation session, the participants completed a logical reasoning task based on Raven's Progressive Matrices during magnetoencephalography. Significant neural responses at the sensor level across all stimulation conditions were imaged using a beamformer. Whole-brain, spectrally constrained functional connectivity was then computed to examine the network-level activity. Behaviourally, we found that participants were significantly more accurate following left compared to right parietal stimulation. Regarding neural findings, we found significant HD-tDCS montage-related effects in brain networks thought to be critical for P-FIT, including parieto-occipital, fronto-occipital, fronto-parietal and occipito-cerebellar connectivity during task performance. In conclusion, our findings showed that left parietal stimulation improved abstract reasoning abilities relative to right parietal stimulation and support both P-FIT and the neural efficiency hypothesis. KEY POINTS: Abstract reasoning is a critical component of fluid intelligence and is known to be served by multispectral oscillatory activity in the fronto-parietal cortices. Recent studies have aimed to improve abstract reasoning abilities and fluid intelligence overall through behavioural training, but the results have been mixed. High-definition transcranial direct-current stimulation (HD-tDCS) applied to the parietal cortices modulated task performance and neural oscillations during abstract reasoning. Left parietal stimulation resulted in increased accuracy and decreased functional connectivity between occipital regions and frontal, parietal, and cerebellar regions. Future studies should investigate whether HD-tDCS alters abstract reasoning abilities in those who exhibit declines in performance, such as healthy ageing populations.

Grandiose narcissism associates with higher cognitive performance under stress through more efficient attention distribution: An eye-tracking study.

Narcissism is a part of the Dark Triad that consists also of the traits of Machiavellianism and psychopathy. Two main types of narcissism exist: grandiose and vulnerable narcissism. Being a Dark Triad trait, narcissism is typically associated with negative outcomes. However, recent research suggests that at least the grandiose type may be linked (directly or indirectly) to positive outcomes including lower levels of psychopathology, higher school grades in adolescents, deeper and more strategic learning in university students and higher cognitive performance in experimental settings. The current pre-registered, quasi-experimental study implemented eye-tracking to assess whether grandiose narcissism indirectly predicts cognitive performance through wider distribution of attention on the Raven's Progressive Matrices task. Fifty-four adults completed measures of the Dark Triad, self-esteem and psychopathology. Eight months to one year later, participants completed the Raven's, while their eye-movements were monitored during high stress conditions. When controlling for previous levels of psychopathology, grandiose narcissism predicted higher Raven's scores indirectly, through increased variability in the number of fixations across trials. These findings suggest that grandiose narcissism predicts higher cognitive performance, at least in experimental settings, and call for further research to understand the implications of this seemingly dark trait for performance across various settings.

Human-level few-shot concept induction through minimax entropy learning.

Humans learn concepts both from labeled supervision and by unsupervised observation of patterns, a process machines are being taught to mimic by training on large annotated datasets-a method quite different from the human pathway, wherein few examples with no supervision suffice to induce an unfamiliar relational concept. We introduce a computational model designed to emulate human inductive reasoning on abstract reasoning tasks, such as those in IQ tests, using a minimax entropy approach. This method combines identifying the most effective constraints on data via minimum entropy with determining the best combination of them via maximum entropy. Our model, which applies this unsupervised technique, induces concepts from just one instance, reaching human-level performance on tasks of Raven's Progressive Matrices (RPM), Machine Number Sense (MNS), and Odd-One-Out (O3). These results demonstrate the potential of minimax entropy learning for enabling machines to learn relational concepts efficiently with minimal input.

Response to Difficulty Drives Variation in IQ Test Performance.

In a large (N = 300), pre-registered experiment and data analysis model, we find that individual variation in overall performance on Raven's Progressive Matrices is substantially driven by differential strategizing in the face of difficulty. Some participants choose to spend more time on hard problems while others choose to spend less and these differences explain about 42% of the variance in overall performance. In a data analysis jointly predicting participants' reaction times and accuracy on each item, we find that the Raven's task captures at most half of participants' variation in time-controlled ability (48%) down to almost none (3%), depending on which notion of ability is assumed. Our results highlight the role that confounding factors such as motivation play in explaining individuals' differential performance in IQ testing.

Learning Visual Abstract Reasoning through Dual-Stream Networks

Visual abstract reasoning tasks present challenges for deep neural networks, exposing limitations in their capabilities. In this work, we present a neural network model that addresses the challenges posed by Raven’s Progressive Matrices (RPM). Inspired by the two-stream hypothesis of visual processing, we introduce the Dual-stream Reasoning Network (DRNet), which utilizes two parallel branches to capture image features. On top of the two streams, a reasoning module first learns to merge the high-level features of the same image. Then, it employs a rule extractor to handle combinations involving the eight context images and each candidate image, extracting discrete abstract rules and utilizing an multilayer perceptron (MLP) to make predictions. Empirical results demonstrate that the proposed DRNet achieves state-of-the-art average performance across multiple RPM benchmarks. Furthermore, DRNet demonstrates robust generalization capabilities, even extending to various out-of-distribution scenarios. The dual streams within DRNet serve distinct functions by addressing local or spatial information. They are then integrated into the reasoning module, leveraging abstract rules to facilitate the execution of visual reasoning tasks. These findings indicate that the dual-stream architecture could play a crucial role in visual abstract reasoning.