perceptual tuning of a simple box sonit bafna anna losonczi john peponis georgia institute of...

Perceptual tuning of a Simple box

Sonit BafnaAnna LosoncziJohn Peponis

Georgia Institute of TechnologyCollege of Architecture, Atlanta Georgia

Space Syntax Symposium 8 Santiago Chile 2012

Proposition

Architectural interest is aroused when a setting is able to inspire a richness and variety of percepts and alternative visual and spatial interpretations.

In a carefully designed space, even subtle changes in location can lead to a rich and meaningful variation in perception.

CasePulitzer Foundation for the ArtsSt. Louis, Missouri

Tadao Ando, Pulitzer Foundation for the Arts, St Louis, Missouri

Richard Serra, Joe

Tadao Ando, Pulitzer Foundation for the Arts, St Louis, Missouri

Tadao Ando, Pulitzer Foundation for the Arts, St Louis, Missouri

The Experimental Set-Up

Hypothesis

The same space visualized from a set of photographs taken from different vantage points will be described in different ways depending on the vantage points.

Selection of vantage points

1 2

3

5

4

Views from the five vantage points were combined to produce four paths, each made of a combination of three views.

Paths

A 135B 145C 235D 245

Paths

A 135B 145C 235D 245

Selection of Subjects

Three categorical variables (2 control)Path: A / B / C / DSex: Female (F) / Male (M)Background: Design (V) / Non-Design (N)

(4 X 4) + 2 = 18 subjects / path4 path (A, B, C, D) 18 X 4 path = 72 subjects72 subjects X 5 = 360 sentences

Data

Subject response sheet

Subjects’ verbal responses were restricted to a specific format offering a choice of four verbs; each response contained at least one prepositional phrase relating the verb to an element or area within the given view.

Sample of compiled data with some basic analysis

Sample of coded data showing counts of relations between elements (used for result #2 below)

1 2

3

5

4

Numbers of prepositions appearing in sentences for each path

Results

0

1

2

3

4

5

6

Pre

pPhr

ases

Spa

tial

A B C D

Path

The subjects assigned to path D produced sentences with more prepositional phrases compared to those assigned other paths. (For instance: I am standing in the middle of the pathway along the window that I am looking through and I see the water surrounding.)

1. Prepositional phrases per sentence

1 2

3

5

4

F=7.57N=360P < 0.001

Oneway Analysis of PrepPhrasesSpatial By Path (PrepPhrasesSpatial refers to number of prepositional phrases per sentence) Oneway Anova Summary of Fit Rsquare 0,060019 Adj Rsquare 0,052098 Root Mean Square Error 0,998094 Mean of Response 1,655556 Observations (or Sum Wgts) 360 Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Prob > F Path 3 22,64444 7,54815 7,5770 <,0001* Error 356 354,64444 0,99619 C. Total 359 377,28889 Means for Oneway Anova Level Number Mean Std Error Lower 95% Upper 95% A 90 1,43333 0,10521 1,2264 1,6402 B 90 1,51111 0,10521 1,3042 1,7180 C 90 1,60000 0,10521 1,3931 1,8069 D 90 2,07778 0,10521 1,8709 2,2847 Std Error uses a pooled estimate of error variance

Response PrepPhrasesSpatial Summary of Fit RSquare 0,092125 RSquare Adj 0,079301 Root Mean Square Error 0,983668 Mean of Response 1,655556 Observations (or Sum Wgts) 360 Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Model 5 34,75758 6,95152 7,1843 Error 354 342,53131 0,96760 Prob > F C. Total 359 377,28889 <,0001* Parameter Estimates Term Estimate Std Error t Ratio Prob>|t| Intercept 1,6590909 0,051925 31,95 <,0001* Path[A] -0,218687 0,089843 -2,43 0,0154* Path[B] -0,155051 0,090217 -1,72 0,0866 Path[C] -0,05202 0,089843 -0,58 0,5629 Background[1V] 0,1686869 0,051925 3,25 0,0013* Gender[F] 0,0636364 0,052249 1,22 0,2241 Effect Tests Source Nparm DF Sum of Squares F Ratio Prob > F Path 3 3 22,998742 7,9229 <,0001* Background 1 1 10,211881 10,5538 0,0013* Gender 1 1 1,435354 1,4834 0,2241 Residual by Predicted Plot Prediction profiler

-2

-1

0

1

2

3

4

Pre

pPhr

ases

Spa

tial

Res

idua

l

0 1 2 3 4 5 6

PrepPhrasesSpatial Predicted

01

234

56

Pre

pPhr

ases

Spa

tial

1.54

5455

±0.2

4778

7

A B C D

A

Path

1V 2N

1V

Background

F M

M

Gender

0

0,5

1

1,5

2

2,5

3

Diff

useA

ttent

ion

A B C D

Path

Each Pair

Student's t

0,05

In comparison to other paths, B elicited many more sentences with diffuse attention; diffuse attention is determined by the presence of prepositional phrases that take as objects broad spatial areas, rather than specific objects.

2. Diffuse versus focal attention

1 2

3

5

4

F=3.06N=360P = 0.0283

Oneway Analysis of DiffuseAttention By Path (DiffuseAttention refers to the number of sentences with prepositional phrases containing a spatial entity) Oneway Anova Summary of Fit Rsquare 0,02514 Adj Rsquare 0,016925 Root Mean Square Error 0,814947 Mean of Response 0,563889 Observations (or Sum Wgts) 360 Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Prob > F Path 3 6,09722 2,03241 3,0602 0,0283* Error 356 236,43333 0,66414 C. Total 359 242,53056 Means for Oneway Anova Level Number Mean Std Error Lower 95% Upper 95% A 90 0,433333 0,08590 0,26439 0,60227 B 90 0,755556 0,08590 0,58661 0,92450 C 90 0,455556 0,08590 0,28661 0,62450 D 90 0,611111 0,08590 0,44217 0,78005 Std Error uses a pooled estimate of error variance

0

1

2

3

4

HO

RIZ

ON

T n

umbe

r

A B C D

Path

Path D is associated with better distributed attention across the entire perceptual field; subjects assigned to path D picked out objects across field of view in a higher proportion of sentences

3. Distribution of attention across field1 2

3

5

4

F=3.65N=360P =0.0168

Oneway Analysis of HORIZONT number By Path (HORIZONT number refers to number of subjects reporting relations between elements across their fields of view) Oneway Anova Summary of Fit Rsquare 0,138556 Adj Rsquare 0,100551 Root Mean Square Error 0,897982 Mean of Response 0,680556 Observations (or Sum Wgts) 72 Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Prob > F Path 3 8,819444 2,93981 3,6457 0,0168* Error 68 54,833333 0,80637 C. Total 71 63,652778 Means for Oneway Anova Level Number Mean Std Error Lower 95% Upper 95% A 18 0,38889 0,21166 -0,0335 0,8112 B 18 0,55556 0,21166 0,1332 0,9779 C 18 0,50000 0,21166 0,0776 0,9224 D 18 1,27778 0,21166 0,8554 1,7001 Std Error uses a pooled estimate of error variance

Discussion

What seems to have distinguished the paths is not so much the overall geometry of the paths, but the difference in the overall complexity of the spatial map that each path supported.

It is not just the information present in each view by itself, but rather the relation between the information provided in each view that influences which element will primarily earn the viewer’s attention; the elements reported in each view carried differential informational content…

… Given that so little discernible detail is available of the window and pool behind it, from vantage point 2, we hypothesized that subjects on path D (vantage points 2 and 4) would tend to report the window and pool much more than those on path B (vantage points 1 and 4).

On path D (245) 17 subjects out of 18 reported the water and window, whereas only 10 subjects of the 18 assigned to path B (145), reported these elements.

1 2

3

5

4

First, our experiments have thrown some light on how cognition is contingent upon a structure of experience; or, put differently, how the structure of a retrieved description is contingent upon the spatial structure of experience within a constant objective structure of space.

Second, we have perhaps found some insight into how subtle design can activate alternative modes of experience and attention, leading to rich descriptions.

Sonit BafnaAnna LosoncziJohn Peponis

Georgia Institute of TechnologyCollege of Architecture, Atlanta Georgia