MISSING CONTROL

Observers Fail to Detect That Behavior is the Control of Perception

A Computer Demonstration of Unintended Writing

Warren Mansell, Autumn Curtis, & Silje Zink

School of Health Sciences, University of Manchester, Manchester, UK

RUNNING HEAD: Missing Control

Correspondence concerning this article should be addressed to:

Reader in Clinical Psychology

CeNTrUM (Centre for New Treatments and Understanding in Mental Health), Division of

Psychology and Mental Health, School of Health Sciences, Faculty of Biology

Medicine and Health, University of Manchester, Manchester Academic Health Science

Centre, 2nd Floor Zochonis Building, Brunswick Street, Manchester, M13 9PL, UK

Tel: +44 (0)161 306 0400; Fax: +44 (0)161 306 0406.

Email address: warren.mansell@manchester.ac.uk

The authors declare no conflicts of interest with respect to this publication.

This research received no specific grant from any funding agency, commercial or not-for-profit

sectors.

The authors have abided by the Ethical Principles of Psychologists and Code of Conduct as set out

by the APA.

© 2019, American Psychological Association. This paper is not the copy of record and may

not exactly replicate the final, authoritative version of the article. Please do not copy or cite

without authors' permission. The final article will be available, upon publication, via its DOI:

10.1037/xge0000590

1

MISSING CONTROL

Abstract

Perceptual control theory (PCT) approaches the behavior of living systems as though it were

a phenomenon of control, and systematically assesses the variables that the individual controls

using the Test for the Controlled Variable (TCV). PCT may be supported by the minority because

the majority of behavior scientists, like most people, can miss the phenomenon of control as it is

occurring. An earlier paper reported three studies of a behavior that was known to be a process of

control because it had been explicitly instructed. In each case, most observers did not detect the

control. Our novel extension of this study used live observation of ‘actors’ and ‘observers’. We

tested in pairs 164 participants randomly allocated to each role. The actors completed a two-

dimensional compensatory tracking task. To keep a dot at the center of a circle, the movements of a

computer mouse needed to vary as the inverse of a disturbance pattern that was an inverted form of

the word "hello" in script. The trace of their mouse movements was displayed on the screen –

writing the word ‘hello’. As predicted, most observers missed the phenomenon of control; they

inferred that the actor’s instruction had been to write ‘hello’, rather than to control the dot. In

contrast, the actors reported that they had been keeping the dot in the circle and were unaware of

having written the word. The TCV analyzes behavior by consistently identifying the controlled

variable without relying on heuristic methods used by researchers that can be inaccurate.

Keywords: goals; closed-loop, intentional inference; self-regulation

2

MISSING CONTROL

Observers Fail to Detect When Behavior is the Control of Perception

A Computer Demonstration of Unintended Writing

The quest to understand the nature of behavior is possibly one of the most enduring, and

most important, challenges for the life and social sciences. In the 1950s, William T. Powers,

proposed a theory of behavior informed by his work in engineering (Powers, Clark, & McFarland,

1960a, 1960b; Powers, 1973, 1978, 2008). Powers observed that living organisms are characterized

not by their ability to emit behavior, but by their ability to control. The term ‘control’ has had no

agreed definition within the field of psychology (Mansell & Marken, 2015; Skinner, 1996). Yet

Powers used one consistent, operational definition: the achievement and maintenance of a desired

state of a variable through actions that counteract disturbances to that variable. Powers constructed

a working model of how control works in living organisms. In his theory, a variable aspect of the

body or environment, such as one’s distance from another person whilst in a conversation, is

compared against a desired reference value for this variable. This reference value is a neural

specification of the desired state for the variable. Actions are continuously varied to counteract

disturbances in the environment to keep this variable aligned with the reference value. So, for

example, if one’s conversation partner moves closer than is preferred, one would move away to

maintain the preferred distance. PCT claims that the only way to understand a person's behavior is

to determine the perceivable aspects of the body and the environment - the perceptual

variables - that the person is controlling (Marken, 1997; Marken & Mansell, 2013). Critically, the

perceptual variables that are controlled are the inputs to the organism, rather than its outputs.

In order to understand behavior in terms of the perceptual variables an organism is

controlling, it is necessary to do research aimed at determining what variables an organism is

controlling. The methods for doing this kind of research are collectively known as such "the test for

the controlled variable" (TCV; Marken, 1997). The TCV requires an individual to make a

hypothesis regarding the nature of the controlled variable and to introduce various disturbances to

3

MISSING CONTROL

test whether the actor keeps the controlled variable constant by counteracting these disturbances.

Since a landmark empirical paper demonstrating this method (Powers, 1978), a series of studies

have demonstrated the accuracy of the TCV across a range of experimental and applied contexts

(for a review, see Marken & Mansell, 2013).

Research using the TCV is critical for a number of reasons. First, it validates the view that

behavior can be accurately conceptualized and measured as control. Second, it allows the researcher

to carry out robust hypothesis testing by building a computational model of individual performance

that can be examined directly for its match with the measured performance (for a review, see

Mansell & Huddy, 2018). Sophisticated versions of these models allow the experimenter to infer a

hierarchy of controlled variables in order to model skills of greater complexity (e.g. Marken, 1986;

Powers, 2008). Third, the TCV provides a scheme to code behavior that is of practical use. Some

examples include the analysis of human sporting performance (Marken, 2001), task analysis for

occupational therapies, rehabilitation, skills training & robotics (Marken, 1999), the provision of

verbal instructions to improve skill performance (Brown-Ojeda & Mansell, 2017), education

(Carey, 2012), and psychological interventions within institutional settings where verbal reports are

limited (Mansell, Carey, & Tai, 2012).

Given the advantages of the Powers’ approach to understanding behavior, it seems

surprising that it has not been adopted more widely. Yet, Powers (1978) showed that what is often

most noticeable about controlling are the irrelevant side effects of the process of control. These side

effects are typically ‘eye-catching’ characteristics of the means used to protect a controlled variable

from disturbance. Powers (1978) demonstrated that researchers often take these side effects as

evidence regarding how behavior works: “A general non-linear quasi-static analysis of relationships

between an organism and its environment shows that the classical stimulus-response, stimulus-

organism-response, or antecedent-consequent analyses of behavioral organization are special cases,

a far more likely case being a control system type of relationship.” (p417, Powers, 1978). For

4

MISSING CONTROL

example, within a standard shock-avoidance task, the most ‘eye-catching’ feature of the experiment

is the frequency of lever pressing by the rat. This is generally regarded as a learned behavior rather

than a side effect of control. However, Powers (1971) used a PCT model to demonstrate that the rat

was varying its lever press rate dynamically to maintain an internally specified probability of its

experience of an electric shock. Powers’ (1971) model showed consistency across individual rats,

indicating that the control of shock probability was a valid way to characterize, and model the rats’

behavior on this task.

Another example of such a traditional analysis is behavioral avoidance, which is a focus of

research within animal behavior (e.g. Löw, Weymar, & Hamm, 2015) and clinical psychology (e.g.

Castagna, Davis, & Lily, 2017). Avoidance behavior is typically assessed as movement away from a

stimulus. For example, research on avoidance of another person is concerned with measuring the

time to respond with avoidance to the approach of the other person and the individual

characteristics, such as gender, that might affect this (Pfaff & Cinelli, 2018). Yet, if we understand

behavior as control, it is apparent that avoidance is merely one instance of an ongoing action that

controls the perceived distance from the approaching person. Indeed the TCV can be used to infer

the desired perceived distance (Bell & Pellis, 2011), and the accuracy of this inference can be

examined by building and testing a computational model against behavioral data (Mansell &

Huddy, 2018). Additional research suggests that an array of other widely researched behaviors can

be reclassified as side effects of control, including fixed action patterns (Marken, 2002), reaction

time in a stimulus detection task (Marken, 2013), and the power law of movement (Marken &

Shaffer, 2017).

In sum, we propose that one reason why controlled variables have not been noticed by

behavioral scientists is because the observable side effects of controlling these variables are often

more compelling – in the sense that they are consistent with researchers’ existing assumptions about

how behavior works – and this makes it difficult to notice that variables are under control.

5

MISSING CONTROL

An earlier series of three experiments (Willett, Marken, Parker, & Mansell, 2017) tested this

proposal. These studies provided all the information that the participants would have needed to

identify the process of control, including making the disturbance to control visible to the observer.

Yet the majority of participants did not detect the control that was occurring. The study concluded

that this effect may explain why the majority of researchers and practitioners miss the variables

under control in studies. However, we wished to examine whether the effect could be replicated in

different conditions, and in doing so, there were some limitations of the earlier research that we set

to address.

The earlier studies had utilized PCT to create a video of a behavior for which the controlled

variable was known, and the actions required to counteract a disturbance left a clear visible trace

(Willett et al., 2017). Specifically, the video showed the hands of the experimenter and a volunteer

who could be seen to each draw a trace using a pen on a whiteboard. Their pens were connected by

a rubber band that had a knot in the middle. The controlled aspect of the environment in this study

was the distance of a knot in the centre of a rubber band from a dot on a whiteboard. The instruction

given to the volunteer (and therefore the volunteer’s intention) was to keep the knot over the dot.

The experimenter and volunteer each held a pen within opposite loops of the rubber band. This

entailed that when the experimenter pulled in one direction, the volunteer needed to pull to the same

degree in the opposite direction to keep the knot over the dot. Thus according to PCT, the

reference value for the volunteer was a distance of zero between the knot and the dot, the

disturbance was provided by the experimenter’s pen movements, and the actions counteracting

this were the movements of the volunteer’s pen. The volunteer’s pen, naturally, left a trace of its

movements on the whiteboard.

Across the three experiments, as predicted, the majority of participants viewing the video

did not report the control that was occurring. Instead they interpreted the trace in a variety of other

ways, for example assuming that the instruction had been “to draw a map of Crete”, or “to draw a

6

MISSING CONTROL

mirror image of the experimenter’s drawing”. The third study confirmed that this effect persisted

even when observer’s attention was directed towards the knot of the rubber band.

There were some aspects of the rubber band study that limited the extent that the findings

might be generalizable. Therefore the current study extended the test. First, the video used in the

original studies only showed a single observer perspective on behavior and it could be argued that

this compromised the judgments. Therefore in the current study the observer viewed live task

performance of the actor trying to keep a dot in the center of a circle on a computer screen using the

movements of a computer mouse. Second, the disturbance in the form of pen movement provided

by the experimenter in the video was spontaneous. In contrast, a pre-specified two-dimensional

disturbance pattern was provided by the computer in the current study and it was concealed from

the observer as often occurs within everyday situations. The experimenter constructed an unseen

disturbance that entailed counteractive action that traced out a recognizable pattern - the word

‘hello’, thereby potentially enhancing the effect.

An additional advantage of live observation not possible in the video version was that the

individual performing the observed task (the ‘actor’) could also be assessed. In particular, one

reviewer of the earlier studies suggested that despite being explicitly instructed to keep the knot

over the dot, the actor may have instead shifted to the goal of doing exactly what the observers had

inferred - such as to draw a pattern, or do the opposite of the experimenter. The current study

allowed this to be examined directly and we anticipated that the actors would maintain their original

instructed goal rather than shifting to any new goals. We also explored the actor’s awareness of the

pattern of movements they had made. Within PCT, there is no requirement for people to be aware

of their actions in order to use them for control.

In sum, we predicted that the majority of observers would not detect that the actor was

controlling the location of the dot to keep it inside the circle, but would instead report the irrelevant

side effect of this control – to write the word ‘hello’ on the screen. In contrast, the actors would

7

MISSING CONTROL

keep to their instruction of keeping the dot in the circle on the screen. Because of its importance in

creating large self-other discrepancies in the understanding of behavior, we also explored the actors’

degree of awareness that they had written the word ‘hello’.

1. Method

1.1. Design

Two participants took part in the experiment simultaneously. They were randomly allocated

using a computerized spreadsheet to one of two Conditions (pattern of movements visible vs

invisible) and to one of two Roles (actor vs observer). We report the first of three trials here and

only those measures relevant to the hypotheses. Ethical permission was granted from the university

research ethics committee.

1.2. Participants

Of 172 recruited participants, three actors in the visible condition did not follow the

instructions and so their data, along with their paired observer, were excluded. A further participant

completed trial 1 but did not complete trials 2 and 3, and so their data, and their paired observer,

were also excluded. Thus, 164 participants were included for the analysis, 41 within each group. All

participants were asked at the end of the study if they knew the outcomes of the study before taking

part and none revealed knowing them in advance. The participants were 87% female (n = 143). Age

ranged from 18-37 years (M = 19.42, SD = 1.84). A two-way Condition X Role ANOVA revealed

no main effect or interactions involving age (ps >.1).

1.3. Materials

1.3.1.Hardware

A desktop PC computer with mouse and mousepad outputted its display to two separate

screens (34x27cm). The actor’s screen was positioned at a right angle to the observer’s screen. For

8

MISSING CONTROL

both conditions, the observer’s screen was partially obscured during the task using a thick black

cloth. The cloth covered the top 2cm and the entire width of the screen to obscure the command

functions on the application. For the trace-invisible condition, the actor’s screen was partially

obscured using a thick black cloth. This cloth covered the top 15cm of the screen and spanned the

entire width. A separate laptop was used to administer questionnaires to the observers.

1.3.2.The Challenger Task (Powers, 2010)

This two-dimensional tracking task was programmed as an executive file on the PC. A circle

of 3mm diameter is displayed at the bottom left of the screen. A dot of 2mm diameter begins at the

centre of the circle. When the trial starts, the dot begins to move in a direction specified by a two-

dimensional disturbance pattern that is the mirror image of the word ‘hello’. The mouse movement

in each combination is additively combined with the disturbance pattern to derive the current

location of the dot at each iteration of the program. The movement of the mouse is instantaneously

displayed at the top left of the screen. A trial lasted two minutes, at which point the movements of

the mouse fully display the word ‘hello’ (6cm in height) on the screen, presuming accurate

performance on the task. Figure 1 displays the block diagram of the Challenger Task based on

perceptual control theory.

9

MISSING CONTROL

Figure 1. A block diagram of the Challenger Task. The dotted line represents the boundary between

the human participant and the environment. The participant perceives the current position of the dot

and compares this with its desired position at the center of the circle. This discrepancy (multiplied

by a gain and coordinated via lower level control systems) drives muscle forces that act against

various unseen disturbances in the environment to produce mouse movements that, in turn,

counteract the disturbance to dot position that is programmed into the computer. The trace of the

actor’s mouse movements writes the word ‘hello’ on the screen. A replica of the actor’s screen is

also viewed by the observer.

10

MISSING CONTROL 11

MISSING CONTROL

1.3.3.Computerized Questions.

Questions were presented using Select Survey (SelectSurvey.NET) to participants on a

computer screen one at a time.

1.3.3.1. Judgment of the Actor’s Instruction. This was assessed by a free text

answer to the question “What were you doing on this task?” for actors, and “What

was the other participant doing in this task?” This was coded as either correct

(“keeping the dot inside the circle” or equivalent) or incorrect (any answer not

describing keeping the dot inside the circle).

1.3.3.2. Judgment of Movements. This was assessed by a free text answer to the

question “What movements was the computer mouse making?” for both actors and

observers. This was coded as either correct (“the word ‘hello’” or equivalent) or

incorrect (any answer not describing producing the word ‘hello’).

1.4. Procedure

Participants were given their instructions in separate rooms. The actor’s instructions were

given in the testing room:

‘Your task in the following minutes will be to use the mouse to control the position of a

small green circle in the bottom left-hand corner of the computer screen. You will need

to try to keep the green circle within the red circle. At the completion of this task I will

ask you to close your eyes. After which you will complete a questionnaire before

completing the task again. This will repeat three times. The task will start as soon as I

hand you the mouse so please be ready to start. Please do not talk until the end of the

study. Do you have any questions?’

The observer’s instructions were given in a private antechamber outside the testing room and were:

‘Your task in the following minutes will be to watch the actor (the other person) and the

computer screen and try to work out what they are trying to achieve. At the completion

12

MISSING CONTROL

of this task, you will be asked to close your eyes. Then you will complete a

questionnaire on a separate laptop before the task is repeated. This will happen three

times. Please do not talk until the end of the study. Do you have any questions?’

After being given their instructions, participants were permitted to ask any questions relating

to their task in the experiment before being reminded not to talk. The observer was then invited into

the room and seated in front of the observer’s screen, at a right angle to the actor and with full view

of the actor’s body but unable to see their screen. The actor was permitted to move the computer

mouse to a more comfortable position. Both participants then filled out the demographics

questionnaire.

Once the questionnaire was completed, participants were asked to close their eyes as the

computer screens were changed, the computer task opened and any covers applied. The actor was

instructed to start immediately after opening their eyes. At the completion of the task the

participants were again asked to close their eyes as the questionnaire was brought up on the screen

and the laptop presented to the observer. During all questions, the observer’s screen was completely

obscured to maintain the confidentiality of the actor’s answers. At the completion of the final

questions, participants were fully debriefed as to the true aims of the research and allowed to ask

any questions before leaving. They were requested not to describe the details of the study to other

students.

1.5. Analysis

In order to test the hypothesis that the observers would be less aware of the actors’

instruction than the actors themselves, we conducted 2 x 2 (Actor-Observer x Correct-Incorrect)

chi-squared tests. To explore discrepancies in awareness of movements, we conducted the tests on

identifications of the pattern of mouse movements as writing the word ‘hello’. All analyses were

completed separately for the trace-invisible and trace-visible conditions to explore whether the

effects occurred when the trace of the word ‘hello’ was visible to the participants.

13

MISSING CONTROL

2. Results

Table 1 reveals that as predicted, only a small number of observers accurately identified the

controlled variable - to keep the dot inside the circle, whereas all actors reported this accurately.

The same effects were found whether the display of the pattern was visible to the actors or not. In

order to gain an understanding of the observers’ judgments, the answers were classified. Out of the

total of eleven observers across both conditions who correctly identified the controlled variable, six

of these had actually elaborated correctly that the actor had moved the mouse in the shape of a word

(‘hello’) in order to keep the dot in the circle. The remaining answers were classified as either ‘to

write ‘hello’ (n = 52), to write a word (n = 6), to draw or trace a pattern (n = 12), or ‘no idea’ (n =

1). Table 2 shows that most actors did not correctly judge that the side effects of their movements

of the computer mouse had spelled out the word ‘hello’, which contrasted with the observers’

judgments, most of whom did identify the correct pattern. Again, the same effects were found

whether the display of the pattern was visible to the actors or not.

14

MISSING CONTROL

Table 1. The frequency of correct and incorrect judgments of the actor’s

instruction by the actor and by the observer, shown separately for each

condition of pattern visibility (visible vs invisible)

Identification of the Instruction

Trace Visibility Role Correct Incorrect χ2 Statistic

VisibleActor 41 0

60.66, p < .001Observer 5 36

InvisibleActor 41 0

57.62, p < .001Observer 6 35

15

MISSING CONTROL

Table 2. The frequency of correct and incorrect judgments of the pattern of the

actor’s mouse movements by the actor and by the observer, shown separately

for each condition of pattern visibility (visible vs invisible)

Identification of movement

pattern

Trace Visibility Role Correct Incorrect χ2 Statistic

VisibleActor 3 38

29.88, p < .001Observer 28 13

InvisibleActor 0 41

39.54, p < .001Observer 28 13

3. Discussion

This study replicated the earlier studies (Willett et al., 2017) showing the observers missed

the controlled variables within a specific instructed task that they observed continuously, and this

study extended the findings to a live observation. These studies support the case that most people do

not immediately consider that behavior is a process of controlling perceived aspects of one’s body

or environment when viewing someone else’s actions. This is evident even when an experiment is

devised in such a way that it is known to be the case that perceptual control is occurring. By

classifying the observers’ answers, it was clear that most of them had assumed that the trace of a

word or pattern on the screen was the actors’ intended action; most of them had identified the word

16

MISSING CONTROL

‘hello’ from the trace. Of the eleven participants (6.7%) who were correct, six of these had actually

realized that the movements tracing the word ‘hello’ were the variable means to control the position

of the dot in the circle on the screen, just as the task had been designed. We did not collect any

further information on how this minority of individuals had managed to correctly identify the

controlled variable.

One possible explanation for the finding that the majority of observers missed the control

that was occurring is that they were basing their judgments on the ‘eye-catching’ features of

behavior - in this case the trace of the word ‘hello’. We propose that behavior researchers may often

use heuristics of this kind, and that these heuristics can obscure the control that is occurring.

Heuristics are recognized to be susceptible to biases in accuracy (Gigerenzer & Brighton, 2009),

and to be situation-specific (Kruglanski & Gigerenzer, 2011), despite their potential accuracy.

Indeed, it has been recognized that heuristics can be used successfully to infer the intentions of

observed behavior (Blythe, Todd, & Miller, 1999).

If the heuristics that humans use to understand behavior are normally accurate, and helpful,

then they would go uncorrected and therefore researchers may persist in using them even if they do

not meet the standards of the scientific analysis of behavior. Examples of heuristics that people may

use to describe behavior may include: a noticeable change in behavior (e.g. an eye movement), an

explicit statement (e.g. the person emitting the behavior states openly what it is they are doing), an

obvious trigger in the environment (e.g. a startle to a loud noise), or a likely intended result of the

behavior (e.g. writing a recognizable word). These heuristics, like those for other human

judgements (Kruglanski & Gigerenzer, 2011), are very different from one another and they apply to

different situational and interpersonal contexts. Therefore they cannot be used without a

consideration of their context to inform a systematic, scientific means to code behavior. Yet,

because their constraints are commonly unquestioned, and they may lead to reports about behavior

17

MISSING CONTROL

that are typically agreed to be correct, we argue, therefore, that the potentially unifying

phenomenon of control goes unnoticed.

All the actors in the study reported that they were indeed following the instructions. This

study therefore challenges an alternative explanation for the findings of Willett et al. (2017) that the

actor in the video was actually following a different instruction from the one requiring control (‘to

keep the knot over the dot’), such as to ‘draw a mirror image’. Indeed, even though the actors in the

current study wrote the word ‘hello’, our results would indicate that the actors were not controlling

for this particular perceptual result of their action.

We expect that our results are a direct result of our use of perceptual control theory to

operationalize behavior as the control of perceptual input. It is important to note that whilst it might

be argued that the observers had correctly identified the aims of the experimenter – to make the

actor write the word ‘hello’ – no participants volunteered this conclusion, and the task explicitly

asked the participants to report only what the actor had been instructed to do. It is also important to

note that it is possible for an observer to use PCT to understand behavior, but this needs to be

carried out using the Test for the Controlled Variable (described earlier), rather than by making the

direct interpretation of the actor’s behavior, as our participants appear to have done.

Our finding also raises the question of how often the same phenomenon of ‘missing control’

might happen in everyday life, in research and in professional practice. To take one simple example,

the lively and exaggerated behavior of a child in class may be the way that the child tries to keep

stimulated in a current environment that lacks sources of stimulation (Zentall, 1975). The observer

may instead interpret this behavior as trying to disrupt other children, or as caused by an underlying

disorder, such as ADHD.

We also found that in this experimental paradigm the actor was typically unaware of the

pattern of their own movements that the observer thinks had been instructed, even when the pattern

is displayed for them to see. It would presumably be more detrimental if observers make false

18

MISSING CONTROL

inferences regarding one’s intentions and one is not even aware of the side effects of the actions

that observers are using as evidence for these views. One clinically relevant example of where this

self-other discrepancy is problematic appears to be cerebral palsy. Interviews reveal that people

with this condition are often unaware of how disorganized their movements appear to others,

leading to frequent misunderstandings regarding the kind of help that other people should provide

them (Martiny, 2015).

Context of the Research

The idea for the current study originated as a way to attempt to replicate three earlier studies

using a novel computer demonstration produced by William T. Powers. These studies have used

tasks in which behavior is demonstrably the control of input; the perceived aspects of the

environment that are kept at a desired state through ongoing actions that counteract disturbances to

this variable. In each study, observers typically notice the ‘eye-catching’ features of behavior and

conclude that it is these outputs that are being controlled. These findings complement our research

program that is aimed at hypothesising and testing the controlled variables (TCV) in an activity, and

in turn building computational models of the activity that are scientifically valid and therefore

replicable (Mansell & Huddy, 2018). For example, we have developed models of visual manual

tracking that are maintained at one-week follow-up and show specificity to each individual

participant (Parker, Tyson, Weightman, Abbott, & Mansell, 2017). We have been able to show that

these models are more accurate and account for signal delay when velocity and position information

are integrated as the controlled input variable. We are now expanding these tests to robotic

demonstrations of basic motor skills. Our future ambition is to demonstrate the validity of the TCV

and model building as a reliable and valid alternative to traditional research within experimental

psychology as applied to social, clinical and human performance domains. For example, we have

shown that conceptualising behavior as the control of input improves the accuracy of drawing

19

MISSING CONTROL

images (Brown-Ojeda & Mansell, 2018), and we are currently developing a computerised task to

extend this principle to object interception within sports performance.

5. Authors’ contribution

W. Mansell conceived of the study and co-designed it with S. Zink and A. Curtis, who both

conducted the study together. W. Mansell analyzed and wrote up the study with support and

feedback from S. Zink and A. Curtis. The results of this study have not been presented elsewhere,

but William T. Powers demonstrated the Challenger program at the International Control Systems

Group Conferrnce, held in Manchester, July 2010. A video of this demonstration has been available

at https://youtu.be/L5b1qeZIn6g since the conference.

6. Acknowledgements

Thank you to Maximillian Parker for his help in piloting the study.

20

MISSING CONTROL

7. References

Bell, H. C., & Pellis, S. M. (2011). A cybernetic perspective on food protection in rats: simple rules

can generate complex and adaptable behaviour. Animal Behaviour, 82(4), 659-666.

Blythe, P.W., Todd, P. M., & Miller, G. F. (1999). How motion reveals intention: Categorizing

social interactions. In G. Gigerenzer & P. M. Todd (Eds.), Simple Heuristics that Make Us

Smart. Evolution and Cognition (pp. 257–285). London: Oxford University Press.

Brown-Ojeda, C., & Mansell, W. (2017). Do perceptual instructions lead to enhanced performance

relative to behavioral instructions? Journal of Motor Behavior. DOI:

10.1080/00222895.2017.1341382

Carey, T. A. (2012). Control in the classroom: An adventure in learning and achievement. Living

Control Systems Publishing.

Castagna, P. J., Davis, T. E., & Lilly, M. E. (2017). The behavioral avoidance task with anxious

youth: A review of procedures, properties, and criticisms. Clinical Child and Family

Psychology Review, 20(2), 162-184.

Gigerenzer, G., & Brighton, H. (2009). Homo heuristicus: Why biased minds make better

inferences. Topics in Cognitive Science, 1(1), 107-143.

Kruglanski, A. W., & Gigerenzer, G. (2011). Intuitive and deliberative judgements are based on

common principles. Psychological Review, 118, 97–109.

Horowitz, J., & Patton, J. (2015). I meant to do that: determining the intentions of action in the face

of disturbances. PloS One, 10(9), e0137289.

Löw, A., Weymar, M., & Hamm, A. O. (2015). When threat is near, get out of here: Dynamics of

defensive behavior during freezing and active avoidance. Psychological Science, 26(11),

1706-1716.

Mansell, W., Carey, T. A., & Tai, S. J. (2012). A Transdiagnostic Approach to CBT using Method

of Levels Therapy: Distinctive Features. London, UK: Routledge.

21

MISSING CONTROL

Mansell, W., & Huddy, V. C. (2018). The Assessment and modeling of perceptual control A

transformation in research methodology to address the replication crisis. Review of General

Psychology, 22(3), 305-320. http://dx.doi.org/10.1037/gpr0000147

Marken, R. S. (1986). Perceptual organization of behavior: A hierarchical control model of

coordinated action. Journal of Experimental Psychology: Human Perception and

Performance, 12(3), 267-276.

Mansell, W., & Marken, R. S. (2015). The origins and future of control theory in psychology.

Review of General Psychology, 19(4), 425-430.

Marken, R. S. (1997). The dancer and the dance: Methods in the study of living control systems.

Psychological Methods, 2(4), 436-446.

Marken, R. S. (1999). PERCOLATe. International Journal of Human-Computer Studies, 50(6),

481-487.

Marken, R. S. (2001). Controlled variables: Psychology as the center fielder views it. American

Journal of Psychology, 114(2), 259-282.

Marken, R. S. (2002). Looking at behavior through control theory glasses. Review of General

Psychology, 6(3), 260-270.

Marken, R. S. (2013). Taking purpose into account in experimental psychology: Testing for

controlled variables. Psychological Reports, 112(1), 184-201.

Marken, R. S., & Mansell, W. (2013). Perceptual control as a unifying concept in psychology.

Review of General Psychology, 17(2), 190-195.

Marken, R. S., & Shaffer, D. M. (2017). The power law of movement: an example of a behavioral

illusion. Experimental Brain Research, 235(6), 1835-1842.

Martiny, K. M. (2015). How to develop a phenomenological model of disability. Medical Health

Care and Philosophy, 18, 553-565.

22

MISSING CONTROL

Parker, M., Tyson, S., Weightman, A., Abbott, B., Emsley, R., & Mansell, W. (2017). Perceptual

control models of pursuit manual tracking demonstrate individual specificity and parameter

consistency. Attention, Perception, and Psychophysics, 79(8), 2523-2537. DOI:

10.3758/s13414-017-1398-2

Pfaff, L. M., & Cinelli, M. E. (2018). Avoidance behaviours of young adults during a head-on

collision course with an approaching person. Experimental Brain Research, 236(12), 3169-

3179.

Powers, W. T. (1971). A feedback model for behavior: Application to a rat experiment. Behavioral

Science, 16(6), 558-563.

Powers, W. T. (1973). Behavior: The control of perception. New York, USA: Hawthorne.

Powers, W. T. (2010). The Challenger Task. Demonstrated at the Control Systems Group (CSG)

International Conference, University of Manchester, Manchester, UK, 24th July 2010.

Powers, W. T., Clark, R. K., and McFarland, R. L. (1960a). A general feedback theory of human

behavior. Part I. Perceptual and Motor Skills, 11, 71-88.

Powers, W. T., Clark, R. K., and McFarland, R. L. (1960b). A general feedback theory of human

behavior. Part II. Perceptual and Motor Skills, 11, 309-323.

Skinner, E. A. (1996). A guide to constructs of control. Journal of Personality and Social

Psychology, 71(3), 549-570.

Willett, A., Marken, R. S., Parker, M., & Mansell, W. (2017). Control blindness: Why people can

make incorrect inferences about the intentions of others. Attention, Perception, and

Psychophysics, 79(3), 841-849. [3]. DOI: 10.3758/s13414-016-1268-3.

Zentall, S. (1975). Optimal stimulation as theoretical basis of hyperactivity. American Journal

of Orthopsychiatry, 45(4), 549-563.

23