Deep Residual Networks’ Abstract Reasoning Performance on Raven’s Progressive Matrices

2021 ◽  
Vol 2 (5) ◽  
Author(s):  
Shuyu Wang
2021 ◽  
Vol 12 ◽  
Author(s):  
Tiansheng Xia ◽  
Mengxia Kang ◽  
Meng Chen ◽  
Jia Ouyang ◽  
Fei Hu

Design training programs that teach creativity often emphasize divergent thinking (generation of ideas) more than convergent thinking (evaluation of ideas). We hypothesized that training would lead to more both types of creativity, but especially divergent thinking. Three groups of university students (N=120; n=40 in each group) were recruited to participate: senior design students (graduate students with at least 4years of design training as undergraduates); junior design students (undergraduates in their first year of design training); and undergraduate students in majors unrelated to design. The students completed three tasks in a classroom setting to assess divergent thinking (Alternate Uses Task), convergent thinking (Remote Associates Task), and nonverbal abstract reasoning (Raven’s Progressive Matrices Test). The results of one-way ANOVAs showed that as expected, senior design students significantly outperformed junior design students and non-design majors in divergent thinking. However, contrary to expectations, senior design students had significantly lower scores than the non-design group on convergent thinking; the junior design students’ scores fell in the middle but were not significantly different from either of the other groups. There were no group differences in nonverbal abstract reasoning. These findings suggest that design training significantly improves students’ ability to generate ideas but does not improve, or may even hinder, their ability to evaluate whether the ideas are useful for the task at hand. The results have implications for developing a research-based curriculum in design training programs.


2021 ◽  
Author(s):  
Mary Flaim ◽  
Aaron Blaisdell

Raven’s progressive matrices (RPM) is a nonverbal intelligence test that examines abstract reasoning by asking subjects to correctly complete a stimulus matrix where transformations between stimuli in the matrix follow one or more relational rules. While this test has been used since 1936, and has been modified to accommodate a variety of humans, this is the first adaptation for non-human animals. We trained pigeons on a task in which the complexity of the rules could increase progressively. Pigeons initially were trained on one rule, and if discrimination reached criterion performance, were tested on novel exemplars to assess relational control. After learning one rule, some pigeons were trained on a second rule, followed by transfer tests. We chose pigeons because of their previous success in abstract and relational rule learning. Pigeons varied in how many rules could be learned and in how well they could transfer learning to novel images, indicating that this test can measure individual differences in learning abstract properties such as stimulus relations. One pigeon successfully learned and transferred two relational rules, and was tested on a matrix combining both rules together. In the future, our modified RPM procedure could be used to investigate abstract reasoning across species.


2008 ◽  
Author(s):  
Denise Paneduro ◽  
Maria Kharitonova ◽  
Nicholas J. Cepeda
Keyword(s):  

2019 ◽  
Vol 11 (1) ◽  
Author(s):  
Christina N. Lessov-Schlaggar ◽  
Olga L. del Rosario ◽  
John C. Morris ◽  
Beau M. Ances ◽  
Bradley L. Schlaggar ◽  
...  

Abstract Background Adults with Down syndrome (DS) are at increased risk for Alzheimer disease dementia, and there is a pressing need for the development of assessment instruments that differentiate chronic cognitive impairment, acute neuropsychiatric symptomatology, and dementia in this population of patients. Methods We adapted a widely used instrument, the Clinical Dementia Rating (CDR) Scale, which is a component of the Uniform Data Set used by all federally funded Alzheimer Disease Centers for use in adults with DS, and tested the instrument among 34 DS patients recruited from the community. The participants were assessed using two versions of the modified CDR—a caregiver questionnaire and an in-person interview involving both the caregiver and the DS adult. Assessment also included the Dementia Scale for Down Syndrome (DSDS) and the Raven’s Progressive Matrices to estimate IQ. Results Both modified questionnaire and interview instruments captured a range of cognitive impairments, a majority of which were found to be chronic when accounting for premorbid function. Two individuals in the sample were strongly suspected to have early dementia, both of whom had elevated scores on the modified CDR instruments. Among individuals rated as having no dementia based on the DSDS, about half showed subthreshold impairments on the modified CDR instruments; there was substantial agreement between caregiver questionnaire screening and in-person interview of caregivers and DS adults. Conclusions The modified questionnaire and interview instruments capture a range of impairment in DS adults, including subthreshold symptomatology, and the instruments provide complementary information relevant to the ascertainment of dementia in DS. Decline was seen across all cognitive domains and was generally positively related to age and negatively related to IQ. Most importantly, adjusting instrument scores for chronic, premorbid impairment drastically shifted the distribution toward lower (no impairment) scores.


2011 ◽  
Vol 34 (6) ◽  
pp. 900-918 ◽  
Author(s):  
Jannica Stålnacke ◽  
Ann-Charlotte Smedler

In Sweden, special needs of high-ability individuals have received little attention. For this purpose, adult Swedes with superior general mental ability (GMA; N = 302), defined by an IQ score > 130 on tests of abstract reasoning, answered a questionnaire regarding their views of themselves and their giftedness. The participants also rated their self-theory of intelligence and completed the Sense of Coherence Scale (SOC-13). At large, the participants experienced being different but felt little need to downplay their giftedness to gain social acceptance. Most participants encompassed an entity self-theory of intelligence, while also recognizing that it takes effort to develop one’s ability. The group scored lower ( p < .001) than Swedes in general on the SOC, which may be a reflection of social difficulties associated with being gifted in an egalitarian society. However, it may also indicate that the SOC carries a different meaning for those with superior GMA.


Author(s):  
Angela Duckworth ◽  

In tandem with increases in delay of gratification, the human capacity for abstract reasoning has increased enormously over the past century. This phenomenon is called the Flynn Effect, after the political scientist who discovered it. I first learned about the Flynn Effect in graduate school. I remember thinking it was impossible. How could it be that as a species, we're getting smarter? And not just a little bit smarter. The size of the Flynn Effect is staggering: more than 30 IQ points—the difference between getting an average score on a standard intelligence test versus qualifying as mentally gifted. Gains are comparable in all areas of the United States and, indeed, around the world.


Stroke ◽  
2012 ◽  
Vol 43 (suppl_1) ◽  
Author(s):  
GALIT WEINSTEIN ◽  
Alexa Beiser ◽  
Rhoda Au ◽  
Charles DeCarli ◽  
Philip A Wolf ◽  
...  

Objectives- Parental stroke is related to an increased risk of stroke among the offspring. Vascular related brain changes, however, often occur before clinical stroke and the association of parental history of stroke and structural brain measures and cognition has not been fully explored. We hypothesized that prospectively verified parental stroke will be associated with increased vascular brain injury and poorer cognitive performance. Methods- A total of 1,297 Framingham offspring (mean age: 61 ± 9 years, 54% women) were studied. Of these, 9.9% had prospectively identified stroke in one or both parents before age 65. Volumetric brain MRI measures of total cerebral brain volume (TCBV), regional brain volumes, white matter hyperintensity volume (WMHV), and covert brain infarcts (CBI) and performance on tests of verbal memory, abstract reasoning, verbal learning and visuospatial memory (VRd) were compared for offspring with and without parental history of stroke. All measures were assessed cross-sectionally and longitudinally (mean duration of follow-up was 6.1±1.2 years). We used models adjusted only for age, sex, education and also additionally adjusted for vascular risk factors and for WMHV as an index of subclinical vascular brain injury. GEE models were used to adjust for sibling relationships among offspring. Results- Higher WMHV (β±SE=0.17±0.08;p=0.027) and lower VRd scores (β±SE=-0.80±0.34; p=0.017) at baseline were found in offspring with parental history of stroke. In addition, participants with parental stroke by age 65 years were more likely to be in the highest quintile of increase in WMHV (OR=1.87;p=0.04) as well as worsening executive function (Trails B-A) (OR:1.81;p=0.03). Parental stroke was not associated with total and regional brain volumes or with memory, abstract reasoning and verbal learning. Conclusions- In our community-based sample of middle-aged asymptomatic subjects, the occurrence of parental stroke by age 65 years is associated with higher baseline WMHV and with a more rapid increase in WMHV. Further, parental stroke is also associated with poorer performance on VRd and a decline in executive function. The effects on baseline WMH and VRd were substantial equivalent to 2.8 and 7 years of brain aging, respectively.


Sign in / Sign up

Export Citation Format

Share Document