Space to ReasonSpace to ReasonMarkus KnauffMarkus KnauffUniversity of Gieß[email protected]‐[email protected] giessen.dewww.uni‐giessen.de/cms/cognition
For every complex problem, there is an answer that is clear simple and wronganswer that is clear, simple, and wrong.
H.L. Mencken (1880 ‐ 1956)
I bilit d Cit Pl iImageability and City Planning
• What does the city's form actually mean to the people who live there? What can the citylive there? What can the city planner do to make the city's image more vivid and memorable gto the city dweller? To answer these questions, Lynch, formulates a new criterion‐imageability‐and shows its potential value as a guide for thepotential value as a guide for the building and rebuilding of cities.
© Markus KnauffUCSB – 04/16/10 2UCSB, Feb/23/12
Visual imagery and reasoning
• Visual imagery helps to reason: Huttenlocher(1968), Shaver, Pierson, and Lang (1976), ( 968), Shaver, Pierson, and ang ( 976),Clement and Falmagne (1986), and many othersothers
• Visual imagery does not help: Sternberg g y p g(1980), Richardson (1987), Johnson‐Laird, Byrne and Tabossi (1989) Newstead PollardByrne and Tabossi (1989), Newstead, Pollard, and Griggs (1986), and many others
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 3© Markus Knauff
The core ideaThe core idea ..
h h d l h f• … to re‐examine the orthodox visual theory of reasoning, to reject it, and to propose a spatial h f dtheory of reasoning in its stead.
• … to show that not visual images, but rather the… to show that not visual images, but rather the ability to mentally construct and inspect more abstract spatial representations is critical forabstract spatial representations is critical for reasoning.
h h b f b h i l i• .. to show that by means of behavioral reasoning experiments, experiments using functional MRI, and
i l d li© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12
computational modeling.4
The visual‐impedance hypothesisKnauff & Johnson‐Laird (2002). Memory & Cognition, 30, 363‐371.
• Orthodox hypothesis: visual relations help to construct visual images and thusto construct visual images and thus support the process of reasoning
• Alternative hypothesis: visual relations elicit irrelevant visual images and thuselicit irrelevant visual images and thus impede the process of reasoning visual impedance hypothesis= visual‐impedance hypothesis
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 55© Markus Knauff
Reasoning and imageabilityg g y
• Four sorts of problems:
• visual problems – e g cleaner-dirtier• visual problems – e.g. cleaner-dirtier
• visuo-spatial problems – e.g. above-below
• spatial problems – e.g. to the north-to the south
• control problems – e.g.better-worse
• Participants solved reasoning problems with the four sorts of relations
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 66© Markus Knauff
Response latencies
i l bl
Knauff & Johnson‐Laird (2002). Memory & Cognition, 30, 363‐371.
• visual problems were significantly slower than the otherthan the other problems (Wilcoxontest z = 3.07; p < .002)p )
• No difference between the other sorts ofthe other sorts of problems
• Imageability does notImageability does not help; it even impedes reasoning
visual control visuo-spatial spatial
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 7
g
7© Markus Knauff
Congenitally Blind ReasonersKnauff & May (2006). Quarterly Journal of Experimental Psychology, 59, 16‐177.
• accuracy [rel. frequency of correct responses]
• response latency [in msec]p ]
1 8000
0,9
4000
6000
0,8
2000
4000
0,7visual relations control
l tivisuospatial
l ti
0visual relations control visuospatial
© Markus KnauffUCSB – 04/16/10
relations relations visual relations controlrelations
visuospatialrelations
8© Markus KnauffUCSB, Feb/23/12
Sighted ReasonersKnauff & May (2006). Quarterly Journal of Experimental Psychology, 59, 16‐177.
• accuracy [rel. frequency of correct responses]
• response latency [in msec]p ]
15001
1100
13000,9
700
9000,8
500
700
visual relations control relations visuospatial0,7
visualrelations
controlrelations
visuospatialrelations
© Markus KnauffUCSB – 04/16/10
prelations
9© Markus KnauffUCSB, Feb/23/12
Blindfolded Sighted ReasonersKnauff & May (2006). Quarterly Journal of Experimental Psychology, 59, 16‐177.
• accuracy [rel. frequency of correct responses]
• response latency [in msec]p ]
11700
0,9 1300
1500
0,8
700
900
1100
0,7visual relations control visuospatial
500
700
visual relations control visuospatial
© Markus KnauffUCSB – 04/16/10
relations relationsvisual relations control
relationsvisuospatial
relations
10© Markus KnauffUCSB, Feb/23/12
Replications and Extensionsof the visual impedance effectof the visual impedance effect
• Dyslexia• Bacon, A.M., Handley, S.J., Dennis I. & Newstead, S.E. (2008). Reasoning strategies: the
role of working memory and verbal spatial ability European Journal of Cognitiverole of working memory and verbal‐spatial ability. European Journal of Cognitive Psychology, 20(6), 1065 ‐ 1086.
• Bacon, A.M., Handley, S.J. and McDonald, E.L. (2007). Reasoning and dyslexia: a spatialstrategy may impede reasoning with visually rich information British Journal ofstrategy may impede reasoning with visually rich information. British Journal ofPsychology, 98(1), 79‐92.
• Psychopharmacology/ Bezodiaziones:Psychopharmacology/ Bezodiaziones:• S. Pompéia , G. M. Manzano, M. Pradella‐Hallinan and O. F. A. Bueno (2007). Effects of
lorazepam on deductive reasoning. Psychopharmacology, 527 ‐ 536
• Individual differences• DeLeeuv, K. Hegarty, M (2008). What Diagrams Reveal about Representations in
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12
e eeu , . ega ty, ( 008). at ag a s e ea about ep ese tat o sLinear Reasoning, and How They Help. Diagrams, 89‐102.
Neural Activity During Reasoning andimageability
• Four sorts of problems:
• visual problems e g cleaner dirtier• visual problems – e.g. cleaner-dirtier
• visuo-spatial problems – e.g. above-below
• spatial problems – e.g. to the north-to the southsouth
• control problems – e.g.better-worse
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 1212© Markus Knauff
Activity in ALL problems Knauff et al. (2003). Journal of Cognitive Neuroscience, 15 (4), 559‐573.
Contrast: all problems vs. baseline (p < .001)
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 1313© Markus Knauff
Activity in visual problemsKnauff et al. (2003). Journal of Cognitive Neuroscience, 15 (4), 559‐573.
Contrast: visual problems vs. control problems (p < .001)
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 1414© Markus Knauff
Reasoning and working memoryKnauff et al., (2004). Spatial Cognition & Computation, 4, 167‐189
• Participants solved transitive inferences (N = 3 X 48) as main task (k = 48)
• Four different secondary tasks were solved concurrently:concurrently:
• Visuo-spatial: location of objects (1)
• visual: brightness of Objects (2)
ti l l ti f t (3)• spatial: location of tones (3)
• Kontrolle: pitch of tones (4)
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 15
Kontrolle: pitch of tones (4)
15© Markus Knauff
Problems
The red circle lies tothe left of the green rectanglethe left of the green rectangle.
The green rectangle lies to theThe green rectangle lies to the left of the blue square.
Does it follow:
The red circle lies to theThe red circle lies to theleft of the blue square?
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 1616© Markus Knauff
Valid InferencesKnauff et al., (2004). Spatial Cognition & Computation, 4, 167‐189
85,883,3
785
85
90
)
• Visual secondary tasks had no negative effect on
baselinebaseline78,5
674
75
80
es (m
ain
task
)
baseline reasoning performance
• Spatial secondary tasks had a disrupting effect on67,4
65
70
rect
resp
onse had a disrupting effect on
reasoning performance (Wilcoxon test z = 2.19; p < 05)
55
60
% c
or < .05)
• Reasoning and spatial secondary tasks are
50visual acoustical
nonspatial spatial (secondary task)
secondary tasks are processes in the same spatial subsystem of working memory
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 17
non-spatial spatial (secondary task) working memory
17© Markus Knauff
Three phases of an inference…Fangmeier, Knauff, Ruff, & Sloutsky (2006). Journal of Cognitive Neuroscience, 4, 559‐573.
• premise processing phase: comprehension and processing of the premisesprocessing of the premises
• integration phase: construction of a single gintegrated model of the premise information; the premises are no longer p grepresented as separate entities in working memory
• lid ti h l ti• validation phase: evaluation of the logical validity of a presented conclusion
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 18© Markus Knauff
“Visual Areas” only in Phase I and IIFangmeier, Knauff, Ruff, & Sloutsky (2006). Journal of Cognitive Neuroscience, 4, 559‐573.
• premise processing phase:middle occipital gyrus, and superior temporal gyrussuperior temporal gyrus, bilaterally
integration phase: anteriorintegration phase: anterior prefrontal cortex and occipito‐temporal gyrus
validation phase: prefrontal gyrus, inferior p gy ,parietal lobule (L); precuneus (R)
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 19© Markus Knauff
Acoustical problemsFangmeier & Knauff (2009). Brain Research, 1249, 181‐190.
• premise processing phase: comprehension and processing of the premisesthe premises
• integration phase: construction of i l i d d l f ha single integrated model of the
premise information; the premises are no longer represented asare no longer represented as separate entities in working memory
• validation phase: evaluation of the logical validity of a presented
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12
g y pconclusion
20© Markus Knauff
Acoustical ProblemsFangmeier & Knauff (2009). Brain Research, 1249, 181‐190.
• premise processing phase:superior parietal gyrus,
bil llprecuneus, bilaterally
integration phase: superior integration phase: superior temporal gyrus, anterior heschl gyrus (BA 41 , 42)
APFC
validation phase: prefrontal gyrus (L), parietal p gy ( ), plobule (L); precuneus (R)
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 21© Markus Knauff
Verbalizer vs. Visualizer(Gazzo & Knauff, in prep.)
Easy to visualize difficult to visualize Easy to visualize difficult to visualize
rors
(%)
rs(%
)
err
error
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 22© Markus Knauff
Individual differences and neural activity
P i i (N 12) l d 32 i i I f
Ruff, C. C., Knauff, M., Fangmeier, T., & Spreer, J. (2003). Neuropsychologia, 41, 1241‐1253
• Participants (N = 12) solved 32 transitive Inferences
• participants spatial‐constructive Intelligence were tested with the “Block Design Test” of the German equivalent to the Wechsler Adult Intelligence Scale (HAWIE‐91)
• min: IQBDT = 103
• max: IQBDT = 128QBDT
• mean: IQBDT = 114.
• positive correlation of spatial‐constructive intelligence and number of correct responses (r = .76, p < .01)
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 2323© Markus Knauff
Negative correlation between BOLD and visuo‐spatial IQRuff, C. C., Knauff, M., Fangmeier, T., & Spreer, J. (2003). Neuropsychologia, 41, 1241‐1253
Contrast: all Problems vs. baseline (p < .001) with BDT score as Covariate (x = 0, Y = - 62, z = 36).
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 24
(p ) ( , , )
24© Markus Knauff
Preferred mental models
V ifi ti f t di ll lid d
Rauh, R., Hagen, C., Knauff, M., Kuß, T., Schlieder, C., & Strube, G. (2005). Spatial Cognition & Computation, 5, 239‐26Jahn, G., Knauff, M. & Johnson‐Laird, P. N. (2007). Memory & Cognition, 35, 2075‐2087
• Verification: faster andmore often correctly
60
70
80
90
100
Logically valid, expected
0
10
20
30
40
50
60
1 2 3 4 5 6 7 8 9 10 11 12 13
15 4
20,820
25
Empirically found
100
15,4
7 310
15
50
60
70
80
90
100
5,257,3
0
5
10
0
10
20
30
40
50 0preferred non-preferred
errors in % resonse time in sec
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12
01 2 3 4 5 6 7 8 9 10 11 12 13
errors in % resonse time in sec
25© Markus Knauff
Preferred Conclusions in Reasoning:Berendt’s “visual account”Berendt s visual account
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 26© Markus Knauff
Preferred Conclusions in Reasoning:Schlieder’s “spatial account”Schlieder s spatial account
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 27© Markus Knauff
Modelling results
• How many preferences can be reconstructed?
• I t 18 t f 72 f• Imagery account: 18 out of 72 preferences unexplained
• Spatial account: 2 out of 72 preferences unexplained
• The spatial a o nt is more parsimonio s and an• The spatial account is more parsimonious and can reconstruct more empirically found preferences.
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 28
The PRISM modelRagni, M. Knauff, M., & Nebel, B. (2005). Proceedings of the 27th Annual Conference of the Cognitive Science Society, (pp. 1797‐1802).Ragni, M., Fangmeier, T., Brüssow, S., & Knauff, M. (submitted).
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 29© Markus Knauff
Knauff, M. (to appear). Space to Reason. C b id MA MIT PCambridge, MA: MIT Press.
© Markus KnauffUCSB – 04/16/10 3030© Markus KnauffUCSB, Feb/23/12
T k h
• previous studies have often shown activation of visual
Take‐home‐message
• previous studies have often shown activation of visual association cortices which points to the role of “visual mental imagery” in reasoningg y g
• the theory explains why visual brain areas are indeed involved in premise processing and the construction ofinvolved in premise processing and the construction of an initial static representation of the initial model
• b t th t b t t ti l t ti h ld i• but that more abstract spatial representations held in parietal cortices are crucial for the actual reasoning processesprocesses
• Thus: visual images can impede reasoning
© Markus KnauffUCSB – 04/16/10 © Markus KnauffUCSB, Feb/23/12 3131© Markus Knauff
Space to ReasonSpace to ReasonMarkus KnauffMarkus KnauffUniversität Gieß[email protected]‐[email protected] giessen.dewww.uni‐giessen.de/cms/cognition
For every complex problem, there is an answer that is short simple and wronganswer that is short, simple, and wrong.
H.L. Mencken (1880 ‐ 1956)
Top Related