You are on page 1of 5

BPK 207 – Speed-accuracy trade-off Copyrighted material Simon Fraser University

SECTION 6 – SPEED-ACCURACY TRADE-OFF

6.1 Section Learning Outcomes:


By the end of this section, you will be able to…
1. Compare the three different speed-accuracy tradeoff relationships based on the type of motor
skill

6.2 Speed-Accuracy Trade-off overview


This occurs in situations where the goal is to move a limb (or other body part) as fast as possible to
reach a target or intercept an object with minimal error. Examples include typing, hitting a ball,
moving your foot to brake while driving, and any rapid movement to a target. It is important in many
tasks, ranging from sports to controlling complicated machinery. There are three different paradigms
with specific speed-accuracy trade-off relationships that we will discuss in this course: logarithmic
(related to Fitts’ paradigm), linear, and temporal.

6.3 Fitts’ Paradigm and Law:

6.3.1 Experimental paradigm


A person moves a stylus back and forth to two targets as fast as possible and as accurately as possible.
Movement time (which relates to speed) is measured. The experimenter manipulates target width (W)
and amplitude (A). Here, amplitude is the distance between targets (see right panel of figure below).
The emphasis is on accuracy rather than speed.

From: Schmidt and Lee 2005

By manipulating the target width and amplitude, the experimenter changes the index of difficulty (ID).
The ID is thus related to the movement distance of the limb and to the target width at which it is
aimed:
ID = Log2(2A/W), where A is movement amplitude and W is target width

6.3.2 Logarithmic Speed-accuracy Trade-off (Fitts’ Law):


The relationship between speed and accuracy in this paradigm, when graphed with a logarithmic
function, produces a straight line.
MT = a + b[log2(2A/W)], where: MT = movement time; a and b = constants; A = amplitude of
movement; W = target width or size. Thus: MT = a + b(ID)
1
BPK 207 – Speed-accuracy trade-off Copyrighted material Simon Fraser University

From: Schmidt and Lee 2005

In the figure above, the different symbols represent different target widths (in inches). MT = average
movement time per tap computed from the # of taps in given time period. For example: 50 taps in 20
seconds = 400 ms/tap (or 0.4 s/tap).

6.3.3 Interpreting Fitts’ Law:


Fitts’ Law shows an inverse relationship between ‘difficulty’ of movement and speed. It describes how
MT must be traded-off to maintain accuracy under different values of ID.

The constant ‘a’ is the intercept. Think of it as a case where no accuracy is required (ID = 0). What is an
example of an ID of 0? When two targets overlap in width!

The slope is ‘b’. It represents the added MT caused by increasing the ID. One thing that changes b is
what body part or “effector” is moving. For example, Fitts’ paradigm can be conducted with
movements of only the arm, or only the wrist, or only the fingers. These different effectors each have
their own sensitivity to changes in ID and therefore, different
effectors show different slopes. See the figure on the right. In this
figure, A = arm movements, W = wrist movements, and F = finger
movements. Note how the slope increases for larger effectors.
This means that larger and more cumbersome limbs are more
sensitive to changes in ID. It also means that the fingers can be
controlled more precisely. Older adults often have higher slopes.

Fitts’ Law holds true for different age groups, lower and upper
limb movements, and even underwater and imagined
movements.

From: Schmidt and Lee 2005

2
BPK 207 – Speed-accuracy trade-off Copyrighted material Simon Fraser University

6.4 Linear Speed-Accuracy Trade-off:


This relationship is observed during rapid single-aiming movements. In this type of paradigm, the
person reaches with a stylus from a starting position to a target 10 – 60 cm away. Movement time is
constrained (or specified) by the experimenter. Target width is maintained. The task requires timing
and distance accuracy.

Errors are measured as standard deviation (SD) of movement amplitude, also called the variation in
movement end point (We).

The linear speed-accuracy trade-off relationship is:

We = a + b(A/MT),

where a and b are constants, A = movement amplitude, and MT = movement time.

Look at the figures below. In the task, MT is constrained to one of three values (140, 170, 200 ms)
during a reaching movement.

From: Schmidt and Lee 2005

The figure on the left (above) shows increased variability (We) as movement amplitude (A) increases
(hint: take any line and note the linear increase). It also shows that W e increases for a given movement
amplitude as the MT decreases (hint: take a single movement amplitude (e.g., 10 cm) and note the
increase between the open square, solid circle, and open circle).

The figure on the right is simply the figure on the left re-plotted to show the speed-accuracy trade-off.
How does this work? Recall that Velocity = A/MT. Thus, this relationship means that as the velocity of
the movement increases for a given movement amplitude, the We (or error or variability) increases
(i.e., accuracy decreases).

Why do we see a linear relationship in one case and a logarithmic relationship in the other case?
Differences between paradigms can offer an explanation, which relates to the feedback hypothesis
and movement-time goal hypothesis (see below).

3
BPK 207 – Speed-accuracy trade-off Copyrighted material Simon Fraser University

Comparisons with Fitts’ paradigm:


- Fitts’ paradigm uses a continuous task, whereas here it is a discrete task.
- MT is constrained here but is a dependent variable in Fitts’ task. We is the dependent variable
here.
- In Fitts’ paradigm there are no errors (or they are within 5% margins) – otherwise the trial is
removed and/or you do the condition again.
- The movement amplitude is varied for both.

Feedback hypothesis:
This hypothesis states that the logarithmic trade-off occurs for movements that are controlled by
feedback-based corrections. It also states that linear trade-offs occur for tasks that are entirely
preprogrammed (no feedback used) like the rapid reaching task just described.

Movement time goal hypothesis:


This hypothesis states that single-aiming paradigms use controlled MTs, which encourage participants
to adopt a non-corrective, rapid control strategy. In Fitts’ task the MT goal is to be as fast as possible.
Essentially, the strategies to perform the tasks are different. However, both argue the same thing in
the end: to increase accuracy, move slower.

6.5 Temporal Speed-Accuracy Trade-off:


This relationship relates to tasks that require anticipation and timing, an example of which is hitting a
baseball. You need to anticipate the flight of the ball, internal movement processes (i.e., motor
planning), and limb movement (recall effector anticipation, Section 4.3.1) to decide when to swing.

The figure below shows a typical paradigm to demonstrate this relationship. In this paradigm, you
must move a slider to intercept (or hit) a target moving along a track.

From: Tresilian 2012

In this paradigm, accuracy is measured in terms of errors of time (early/late arrival). Interestingly, the
more ‘violently’ the person performs the movement (i.e., smaller MT or larger movement velocity),
the more accurate the timing. This is opposite to the speed-accuracy principle discussed above for
spatial accuracy related movements.

4
BPK 207 – Speed-accuracy trade-off Copyrighted material Simon Fraser University

In a variation of the above paradigm, you can have a person perform discrete movement-timing tasks.
In this case, the goal is to produce a specific MT. The dependent variable is variable error in timing
(VEt) or SD of the MTs. Note the table below that shows the results of such an experiment.

From: Schmidt and Lee 2005

This table shows that smaller MTs, given the same distance, produce improved movement-timing
consistency. In other words, movement duration is more consistent (VE is lower) with decreasing MT,
or MT consistency is increased as velocity is increased. This relationship holds for discrete but also
repetitive-timing tasks.

Consider the following exercise to demonstrate the above point. Have someone time you trying to
estimate 2 seconds. In other words, have someone tell you when to go, and then you count to 2 in
your head and tell that person when you reach that number. Note whether you are correct. Now try it
when counting to 20 seconds. It is easier to estimate shorter time intervals.

Application:
The temporal speed-accuracy tradeoff suggests that a baseball batter should swing ‘harder’ (with a
smaller MT) so that errors in timing decrease. This could be one possible explanation for why batters
practice swinging a bat with a weight on the end. Once taking the weight off, you swing harder since
you have compensated for the increased weight.

References (and sources):

• Magill RA. Motor learning and control: concepts and applications. 8th Edition. Boston, MA: McGraw-Hill, 2007.
• Schmidt RA, Lee TD. Motor Control and Learning: A Behavioral Emphasis. 4th Edition. Champaign, Ill: Human
Kinetics Publishers, 2005.
• Tresilian J. Sensorimotor control and learning. An Introduction to the Behavioral Neuroscience of Action. Palgrave
Macmillan, 2012.

****All figures are used in accordance with Copyright Act and SFU Copyright guidelines****

You might also like