Hick-Hyman Law

Report
Hick-Hyman Law
INFORMATION THEORY
1.
THE COMMUNICAION SYSTEMS
○
Channel capacity (C) – the amount of info transmitted per
time through a channel
Division of Information
Management Engineering
Hick-Hyman Law
○
the rate of gain of information (Hick, 1952) and index of
performance (IP) in Fitts (1948)
Division of Information
Management Engineering
Hick-Hyman Law
2.
QUANTIFYING INFORMATION
○
○
Information – reduction in uncertainty (bit)
Shannon-Weiner measure of information
=
○
○
○

=1 
log 2
1

or  =

=1 
log 2 
Have : the entropy of a stimulus or a set of stimuli when the
alternatives are not equiprobable
Hmax : the alternatives are equiprobable
HT = H(x) – Hy(x)
where H(x): the expected information of the source
Hy(x): the received information at the destination
Division of Information
Management Engineering
Hick-Hyman Law
THE HICK-HYMAN LAW
1.
Hick (1952) Original Experiments
 choice RT vs. stimulus info content
 errorless responses
Division of Information
Management Engineering
Hick-Hyman Law
○
Experiment II
○
3 phases – as fast as possible, then as accurately as
possible, finally as fast as possible again
training
(accurate)
diamonds for
fast RT
Division of Information
Management Engineering
Hick-Hyman Law
2.
Hyman (1952) Original Experiments
Division of Information
Management Engineering
Hick-Hyman Law
2.
Hyman (1952) Original Experiments
○ The first to articulate the linearity between RT and HT
○ Altered the probabilities of the stimuli to assess RT
as a function of HT
○ RT was linear as s function of bits of the alternatives
with unequal probabilities
○ RT = a + b HT
○
1/b: the rate of gain of information (information capacity)
Division of Information
Management Engineering
Hick-Hyman Law
4.
Research and Applications
Speed-Accuracy Tradeoff
Stimulus-Response Compatibility (SRC)
○ Compatible S-R pairs facilitate the responding of a stimulus,
thus yielding a higher rate of information transfer
Psychometrics
○ investigate RT-IQ relationship
HCI Applications
Division of Information
Management Engineering
Fitts’ Law
5. FITTS’ LAW
○ a linear relationship between task difficulty (ID) and RT
 = log 2
2

○ human motor system as a communication channel, movement
amplitude as the signal, target width as the noise
1. Fitts (1954) Original Experiemtns
○
○
the reciprocal tapping task
Experiment I – metal-tipped stylus (1 oz vs. 1 lb); W from 0.25” to
2”; D from 2 to 16 ”; accuracy was encouraged
Division of Information
Management Engineering
Fitts’ Law
○ IP (index of performance or throughput) = ID/MT the capacity of
the human motor system
 =  +  log 2
2

Division of Information
Management Engineering
○
2.
channel capacity (Shannon’s Theorem 17)
+
 =  log 2

○ W is the bandwidth, P is the signal power and N is the
noise power
Theoretical Development
○
+0.5

+
 log 2

Welford (1960)   =  +  log 2
○ MacKenzie (1992)   =  +
○ Meyer et al. (1988)
  =  + 


○ deterministic iterative-correction model (Crossman and Goodeve,
1983), stochastic optimized-submovement model (Meyer and
colleagues, 1990)
Division of Information
Management Engineering
○
Meyer et al. (1998)   =
submovements
3.


1

where n is number of
Research and Applications
○ kinematics and neurocognitive focus
Speed-Accuracy Tradeoff
Psychometrics
HCI Applications
Pointing.
Angle of Approach.
○ the original Fitts’ paradigm – 1D task
Division of Information
Management Engineering
○ Accot and Zhai (2003) – classical paradigm as AP (pointing with
amplitude constraints); paradigm with height constraints as DP
(pointing with direction constraints)
Division of Information
Management Engineering
Semantic Pointing.
○ both decreasing A and increasing W
Text Entry on Soft Keyboards.
○ text entry on GUI
Navigation.
Division of Information
Management Engineering
Hick-Hyman Law
6. INTEGRATION OF THE LAWS
○
Combine the Hick-Hyman Law and Fitts’ Law
○ Beggs et al. (1972)
○ Fitts’ Law did not hold in the fusion
○ Hoffman and Lim (1997)
○ Home-to-target paradigm with both sequential and
concurrent tasks
○ The sum of the decision and movement time (sequential)
○ Substantial interference (concurrent)
○ Soukoreff and MacKenzie (1995)
○ Unable to fit the data to the model
Division of Information
Management Engineering
Hick-Hyman Law
7. THE HICK-HYMAN LAW AND HCI
○
Common characteristics in both Laws
a. Same analogies based on Shannon and Weaver’s (1949)
b. Same measures such as performance rate and information
capacity
c. Substantial support in research
○ Possible reasons for the lack of momentum in HCI (Laming, 1966)
1. discrepancies between Shannon’s theory and Hick’s analogy
2. Victim for the eviction of the soft sciences by hard sciences
I.
Fitts’ Law has also comparable quantitative components
II. HCI has shifted its focus to include some soft sciences
such as sociology
Division of Information
Management Engineering
Hick-Hyman Law
1.
Difficulty in Application
○
2.
Complexity of Stimuli
○
3.
No need to engage in the complexity of the information theoretic
measures
Multidimensional stimuli for the highly complex interfaces
needed with simple unidimensional stimuli to reduce
confounding
Levels and Types of Performance
○
Fitts’ for somewhat monotonous tasks
Division of Information
Management Engineering

similar documents