ArticlePDF Available

Slot machine near wins: Effects on pause and sensitivity to win ratios

Authors:
  • Ministry of Justice New Zealand

Abstract and Figures

When a near-win outcome occurs on a slot machine, stimuli presented resemble those presented when money is won, but no money is won. Research has shown that gamblers prefer and play for longer on slot machines that present near wins. One explanation for this is that near wins are conditioned reinforcers. If so, near wins would produce longer latencies to the next response than clear losses. Another explanation is that near wins produce frustration; if so, then near wins would produce shorter response latencies. The two current experiments manipulated win ratio across two concurrently available slot machines and also manipulated near win frequency. Latencies were longer following near wins, consistent with near wins functioning as conditioned reinforcers. We also explored the effects of near wins on sensitivity to relative win rate and found that higher rates of near wins were associated with greater sensitivity to relative win frequency, an effect also consistent with near wins as conditioned reinforcers.
Content may be subject to copyright.
Analysis of Gambling Behavior Analysis of Gambling Behavior
Volume 8 Article 1
2014
Slot Machine Near Wins: Effects on Pause and Sensitivity to Win Slot Machine Near Wins: Effects on Pause and Sensitivity to Win
Ratios Ratios
Tadhg E. Daly
Victoria University of Wellington
Gordon Tan
Victoria University of Wellington
Lincoln S. Hely
Victoria University of Wellington
Anne C. Macaskill
Victoria University of Wellington
, anne.macaskill@vuw.ac.nz
David N. Harper
Victoria University of Wellington
See next page for additional authors
Follow this and additional works at: https://repository.stcloudstate.edu/agb
Part of the Applied Behavior Analysis Commons, Clinical Psychology Commons, Experimental
Analysis of Behavior Commons, and the Theory and Philosophy Commons
Recommended Citation Recommended Citation
Daly, Tadhg E.; Tan, Gordon; Hely, Lincoln S.; Macaskill, Anne C.; Harper, David N.; and Hunt, Maree J.
(2014) "Slot Machine Near Wins: Effects on Pause and Sensitivity to Win Ratios,"
Analysis of Gambling
Behavior
: Vol. 8 , Article 1.
Available at: https://repository.stcloudstate.edu/agb/vol8/iss2/1
This Article is brought to you for free and open access by theRepository at St. Cloud State. It has been accepted for
inclusion in Analysis of Gambling Behavior by an authorized editor of theRepository at St. Cloud State. For more
information, please contact tdsteman@stcloudstate.edu.
Slot Machine Near Wins: Effects on Pause and Sensitivity to Win Ratios Slot Machine Near Wins: Effects on Pause and Sensitivity to Win Ratios
Authors Authors
Tadhg E. Daly, Gordon Tan, Lincoln S. Hely, Anne C. Macaskill, David N. Harper, and Maree J. Hunt
This article is available in Analysis of Gambling Behavior: https://repository.stcloudstate.edu/agb/vol8/iss2/1
Analysis of Gambling Behavior
2014, 8, 55-70
Number N Number 2 (Winter, 2014)
55
Slot Machine Near Wins: Effects on Pause and
Sensitivity to Win Ratios
Tadhg E. Daly, Gordon Tan, Lincoln S. Hely, Anne C. Macaskill,
David N. Harper, & Maree J. Hunt
Victoria University of Wellington
When a near-win outcome occurs on a slot machine, stimuli presented resemble those
presented when money is won, but no money is won. Research has shown that gam-
blers prefer and play for longer on slot machines that present near wins. One explana-
tion for this is that near wins are conditioned reinforcers. If so, near wins would pro-
duce longer latencies to the next response than clear losses. Another explanation is
that near wins produce frustration; if so, then near wins would produce shorter re-
sponse latencies. The two current experiments manipulated win ratio across two con-
currently available slot machines and also manipulated near win frequency. Latencies
were longer following near wins, consistent with near wins functioning as conditioned
reinforcers. We also explored the effects of near wins on sensitivity to relative win
rate and found that higher rates of near wins were associated with greater sensitivity
to relative win frequency, an effect also consistent with near wins as conditioned rein-
forcers.
Keywords: gambling, near win, near miss, response latency, generalized matching
law
____________________
People who predominantly gamble with
slot machines develop a pathological profile
faster than gamblers favoring other gambling
activities (Breen & Zimmerman, 2002). This
suggests that features of the gambling medi-
um contribute to the likelihood that an indi-
vidual’s gambling will become problematic.
Slot machines are controlled by payout algo-
rithms with features likely to lead to persistent
and frequent play. For example, all slot ma-
chines use random ratio schedules arranging
intermittent reinforcement schedules that typ-
ically yield high rates of responding and high
resistance to extinction (Ferster & Skinner,
1957; Jenkins & Stanley, 1950). Payout fre-
quency (Dixon, Maclin & Daugherty, 2006),
overall payback rate (Haw, 2008) and the de-
lay between obtaining a win and receiving a
__________
Address all correspondence to:
Anne C. Macaskill
Victoria University of Wellington
PO Box 600
Wellington, New Zealand, 6012
E-mail: anne.macaskill@vuw.ac.nz
payout (Chóliz, 2010) are other features of the
slot machine medium that influence gamblers’
preference for and persistence on a given slot
machine.
The presence of near wins may also in-
fluence preference for (Ladouceur & Giroux,
2006; Dymond & Roche, 2010) and persis-
tence on (e.g. Côte, Caron, Aubert,
Desrochers & Ladouceur, 2003; Kassinove &
Schare, 2001) a given slot machine. A near
win (also called a near miss) is a loss that re-
sembles a win; for example, four matching
symbols constitute a near win when the ma-
chine’s only winning combination is five
matching symbols. Slot machines are pro-
grammed to produce a higher-than-chance
proportion of near-win outcomes (Harrigan,
2007, 2008).
The processes through which near wins
affect gambling behavior have yet to be iden-
tified. Kahneman and Tversky (1982) pro-
posed that situations like near-win outcomes
on slot machines produce more frustration
than other losing outcomes because near wins
1
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
56
SLOT MACHINE NEAR WIN FREQUENCY
make it easier to imagine having received a
win. Loftus and Loftus (1983) suggested that
this particular type of frustration, labeled
cognitive regret” by the authors, might be
eliminated by continuing play. Amsel (1958)
proposed that situation, such as near wins,
that resemble those where rewards have pre-
viously been presented produce a frustration
effect that increases the speed and strength of
ongoing operant behavior, in this case, caus-
ing faster responses to escape frustrating near-
win outcome stimuli. This idea was revisited
by Dixon and colleagues (Dixon, et al., 2011;
Dixon, MacLaren, Jarick, Fugelsang & Harri-
gan, 2013). Dixon et al. (2011) found that
arousal as evidenced by variations in skin
conductance responses and heart rate deceler-
ation measures, was greater following near
wins than other types of losses or actual wins.
They argued that these findings, when consid-
ered in the light of prior research on the psy-
chophysiological effects of frustration, were
consistent with the idea that near wins elicit
frustration. They further proposed that, alt-
hough near wins lead to frustration and thus
have no hedonic value, they negatively rein-
force further play as gamblers seek to escape
the negative arousing effect of these out-
comes. Both Amsel and Dixon suggested that
if near wins create a frustration effect, then
response latencies (time from the outcome
until the next response is made) following
them would be shorter than those following
other losses.
An often-proposed alternative mechanism
(e.g. Griffiths, 1999; Ladouceur & Sevigny,
2002; Peters, Hunt & Harper, 2010; Reid,
1986; Skinner, 1953) through which near
wins might affect gambling is conditioned
reinforcement. Kassinove and Schare (2001)
suggested that, if winning spins are occasion-
ally preceded by near-win spins, the joy and
elation experienced from the win stimuli
would eventually spread to the near win. In
fact, the random schedules arranged by real-
world slot machines do not create the condi-
tions needed to establish near wins as condi-
tioned reinforcers in this way. Although the
pairings described by Kassinove and Schare
likely occur, they would be insufficient to es-
tablish near wins as conditioned reinforcers
because contingency rather than mere conti-
guity is required for Pavlovian conditioning.
This means that in order for pairings of con-
secutive spin outcomes to establish near wins
as conditioned reinforcers it would be neces-
sary for wins to be more likely to occur fol-
lowing near wins than following other losses.
Slot machine outcomes are independent, that
is, the probability of a win is identical follow-
ing every spin and near wins do not signal any
increased probability of a win occurring. Alt-
hough there is a lack of contingency between
near win spin outcomes and win spin out-
comes, there is another portion of the se-
quence of events arranged by real-world slot
machines that does arrange a contingency be-
tween near win outcomes and win outcomes.
This occurs within a winning spin: between
when the gambler presses the spin button and
when the final outcome is presented. That is,
because slot machine reels stop one-by-one
from left-to-right, during every win sequence
near win stimuli are displayed before the final
reel stops spinning, displaying the win stimu-
li. This rapid pairing of near win with win
stimuli during every win sequence is an ideal
sequence of events for establishing near wins
as conditioned reinforcers.
Alternatively, Delfabbro and Winefield
(1999b) and later Peters et al. (2010) suggest-
ed Pavlovian generalization as a more
straightforward process through which near
wins might develop conditioned reinforce-
ment effects. That is, if wins are (conditioned)
reinforcers then stimuli that resemble them
near wins may also become conditioned re-
inforcers through generalization.
If near wins are conditioned reinforcers
as a result of either or both of these processes,
they would be expected to produce longer re-
sponse latencies than other losses. Delfabbro
2
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
57
and Winefield (1999a) recorded participants
playing on real slot machines and found that
response latencies were longer following wins
than losses. Peters et al. (2010) found the
same in a rat model of gambling. This is con-
sistent with findings that reinforcers in gen-
eral produce response latencies or “post-
reinforcement pauses” that are longer than
latencies following other responses (Ferster &
Skinner, 1957).
Previous studies have found inconsistent
effects of near wins on response latencies,
meaning that it is not yet possible to deter-
mine whether near wins primarily increase
persistence of play through conditioned rein-
forcement or through frustration. Dixon and
Schreiber (2004) examined response latencies
following near wins, wins, and losses on a
real slot machine and found much between-
individual variability in the effect of outcome
type on latency length. Dixon et al. (2013) in
a simulation with human participants found
shorter response latencies following near
wins than following other losses. Whereas
Peters et al. (2010) found that rats responding
on a slot machine analog task produced long-
er latencies following near wins than other
losses.
The differences in results across these
studies may partially reflect the species stud-
ied, however, there were several other differ-
ences in these studies that may be relevant
and which point to features of the slot ma-
chine program as determinants of the effects
of near wins on response latencies. One pos-
sible contributor to this variability is the pat-
terns of symbols classified as near wins. Both
frustration and conditioned reinforcement as
explanations for the near win effect suggest
that near wins with outcome sequences that
resemble those presented on win trials for the
longest portion of the sequence would pro-
duce a stronger near win effect. Differences
in the length of reels across studies (Dixon et
al., 2013 used three while Peters, et al., 2010
used five) might therefore account for some
of the observed variability. This is also con-
sistent with Dixon et al.’s finding that only
“classic near wins” (two winning symbols
followed by a different symbol) produced dif-
ferential (shorter) response latencies (alt-
hough Ghezzi, Wilson, & Porter, 2006 found
inconsistent effects of the pattern of symbols
comprising a win on persistence). This may
also explain the variability in Dixon and
Schreiber’s (2004) results as they used several
types of near win but did not control the rates
of each pattern and collapsed across them
when calculating latencies. The current study
used only near wins in which the first four
symbols matched while the fifth differed.
Whether wins are presented during the
session may also affect subjects’ responses to
near-wins resembling them (Ghezzi, et al.,
2006). Dixon et. al. (2013) assessed latencies
following near misses where the first two of
three symbols were the jackpot symbol. Par-
ticipants never experienced Jackpots. If – as
previously suggested near wins obtain rein-
forcing effects during presentation of the win
sequence, the near wins in Dixon et al. would
not have become conditioned reinforcers be-
cause win sequences were never experienced.
Furthermore in some procedures, for example,
Kasinove and Schare, (2001) near wins are
initially presented with wins and then pre-
sented without. In such procedures each near
win presented in the absence of wins would
act as an extinction trial, gradually eliminat-
ing any existing conditioned reinforcement
effects. In the current study, participants ex-
perienced wins as well as related near wins
and clear losses.
In addition, we inserted a behavioral
choice paradigm into each game. A procedure
developed by Davison and Baum (2000) was
used to assess sensitivity to relative win fre-
quency. This involved varying the proportion
of wins allocated to each of two reels across a
series of frequently-changing conditions. Lie,
Harper, and Hunt (2009) successfully used
this procedure to assess sensitivity to win ra-
3
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
58
SLOT MACHINE NEAR WIN FREQUENCY
tios in humans responding for hypothetical
money. In this context, sensitivity refers to the
extent to which individuals allocate their re-
sponses across two alternatives in proportion
to the distribution of reinforcers received
from those two alternatives. The generalized
matching law was used to assess this sensi-
tivity because it separates sensitivity to the
rate of wins from bias toward one of the two
slot machines for some other reason such as
the symbol set used. Such biases are likely
given that people have different histories with
gambling contexts prior to taking part in the
research. It is of interest to investigate the ex-
tent to which people are sensitive to the dis-
tribution of wins because gambling is a con-
text in which people demonstrate an apparent
insensitivity to reinforcement rate in that they
continue to gamble although the mean result
is a loss.
EXPERIMENT 1
In Experiment 1, participants played on
computer-simulated slot machines that pro-
duced no near wins in one session and near
wins on 50% of non-winning trials in another
session. Within each session, relative win fre-
quency was also manipulated across four
conditions in order to assess sensitivity to win
ratios. If near wins affect gambling behavior
via conditioned reinforcement, we expected
response latencies following near wins to be
longer than those following other losses.
Conversely if near wins affect gambling be-
havior via frustration, we expected response
latencies following near wins to be shorter
than those following other losses.
METHOD
Participants
Twenty-nine first year psychology stu-
dents from Victoria University of Wellington
participated voluntarily in partial fulfillment
of a course requirement. Three participants
did not complete the required conditions in
the time allotted for either of the two sessions,
one elected to leave before a session ended,
and another was excluded because of their
high Problem Gambling Severity Index
(PGSI) score (see below). Therefore, we in-
cluded 24 participants in the final experiment.
Apparatus and Materials
Participants completed the PGSI, a nine-
item subscale of the Ferris and Wynne (2001)
Canadian Problem Gambling Index (CPGI).
For each item on the PGSI people respond on
a four point scale ranging from ‘never’ (0) to
‘almost always’ (3). The total PGSI score
ranges from 0 to 27, with a score of 3 or high-
er signifying a potential gambling problem.
None of the 24 participants included scored
above 3. One additional student signed up to
participate and received a score above this
threshold. Therefore they were given an alter-
native non-gambling-related task to complete
and were not included in the study. An ab-
sence of gambling problems was an inclusion
criterion for the current experiment because
of ethical concerns with exposing problem
gamblers to gambling-related stimuli.
Four desktop Dell PC dual-core Penti-
um® computers were arranged in the corner of
a room (two along each wall). Each had a
mouse attached that participants used to make
responses. The slot machine simulations were
programmed in Visual Basic 6®. The sounds
of the slot machines were presented via the
computer speakers.
Procedure
Up to two participants completed the ex-
perimental tasks simultaneously in the testing
room. Participants first completed an in-
formed consent form, and the PGSI. The ex-
perimenter then introduced the slot machine
task, and instructed participants to try to win
as much money as possible, to switch freely
between the two available reels while playing
on each computer, and to move to the next
computer when a message on the screen in-
structed them to do so. The experimenter also
4
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
59
told participants to read the instructions on the
screen, these read:
“This is a slot machine task. You
start with $5. On each spin you can
bet between 10c and 30c and you
can choose whether to play ‘SLOT
1’ or ‘SLOT 2’ (you can freely
switch back and forth between the
two slot lines). When all five pic-
tures in a row match each other
you win 50c for every 10c you bet
(e.g. 10c bet = 50c win, 20c bet =
$1 win etc.). The task will automat-
ically stop after 10 mins of play or
12 wins (whatever happens first).
When the task stops please wait
until told what to do next. Any
questions?”
Participants clicked a button labeled
“Start Task” to advance to the playing screen.
There were two five-symbol slot machines
presented vertically aligned on the playing
screen each with radio buttons displayed to
their right that could be used to select a bet
amount of 10c, 20c or 30c (see Figure 1). The
symbols on each reel were from a visually
distinctive set.
At the start of each trial, participants se-
lected a reel to play, chose an amount to bet,
and then clicked the associated play button in
order to initiate a “spin” on the selected reel.
When this button was clicked, slot machine
spinning sounds played while a slot-machine
animation occurred. During this animation, all
slot stimuli were removed for 150ms and
were then displayed for 150ms creating a
flashing effect. For the first 600ms, different
symbols were presented in every position dur-
ing each flash. After 600ms, the left-most
symbol became fixed, and one additional
symbol became fixed every 300ms until the
five symbols associated with the trial outcome
were presented.
The number of matching symbols from
the left was associated with the outcome of
the spin. Three types of outcome were possi-
ble: win, near win, and clear loss. If a win oc-
curred, five matching symbols were present-
ed, a ringing bell sounded, and participants
saw a message stating that they had won five
times the bet amount (e.g. bet 30c and win
$1.50). Note that money bet and won was hy-
pothetical. On near-win trials the four left-
most symbols matched, and on clear loss tri-
als either two or no matching symbols were
presented (no spin ever resulted in three
matching symbols). On near-win and clear
loss trials no money was won, and partici-
pants saw a message stating that they had won
$0. After each outcome the participant’s cur-
rent “total balance” was updated on screen.
Additionally, after the computer displayed an
outcome, all the screen elements reappeared
and the computer de-selected the bet selection
radio buttons.
The current experiment manipulated two
independent variables in a within subjects, 2 x
4 factorial design producing 8 conditions.
Each condition lasted for 12 wins (obtained
from both slot reels) or 10 minutes, whichever
came first. Of the 192 conditions completed
in Experiment 1 (8 conditions for each of the
24 participants), 154 finished after all 12 wins
were obtained and 38 finished after reaching
the 10-minute time limit.
The first independent variable was the
probability of a near win occurring on a trial
on which a win was not programmed. Wheth-
er a near win was presented was determined
randomly with replacement for each non-win
trial for each participant. During one session
this probability of a near win occurring on a
non-win trial was 0, and in the other 0.5. For
the session including near wins, this arrange-
ment of outcomes meant that there was no
contingency between near wins and wins.
That is, near wins signaled nothing about the
likelihood of a win on the following trial.
Analyses of outcomes actually experienced by
5
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
60 SLOT MACHINE NEAR WIN FREQUENCY
Figure 1. Screenshot showing the play screen. During the play animation, only the chosen row
was visible. The top row depicts a near win. The bottom row depicts a clear loss.
the current participants confirmed that they
were independent in this way. Sessions were
no more than one week apart.
The second independent variable was
the distribution of the 12 wins across the two
reels. Wins were presented according to a
dependently-scheduled variable-interval
(VI) 10 schedule and the proportion of wins
allocated to each reel was manipulated with-
in each of the two sessions. A one-spin
changeover delay was in effect meaning
that, even if a win had been allocated to a
given reel, it was not presented until the
second spin made on that reel following a
switch. The four win distributions were
2:10, 10:2, 4:8, and 8:4, where the first
number indicates the number of wins allo-
cated to the top reel and the second the
number allocated to the bottom reel.
Each of the eight conditions was associ-
ated with a different background screen col-
or and presented on a different computer.
When participants completed a condition,
the computer displayed an end screen
prompting them to move to the next com-
puter to complete the next condition. Twelve
participants completed the conditions in the
order: 2:10, 10:2, 4:8, and 8:4; the remaining
12 completed the conditions in the order:
4:8, 8:4, 2:10, and 10:2. Which near win
frequency participants experienced during
their first session was also counter-balanced.
Neither changes in win distribution within
each session nor changes in the probability
of near wins between sessions were accom-
panied by any additional stimulus changes.
Dependent variables were the proportion of
spins, amount bet, and response latencies for
each reel. The response latency was defined
as the duration between a trial outcome on
trial ‘n’ and the response to initiate the spin
on the subsequent trial ‘n+1’.
RESULTS AND DISCUSSION
We calculated the median response la-
tencies following wins, near wins and clear
losses for each participant for each of the
eight conditions. We averaged the means of
these median response latencies to produce
mean response latencies for each participant
6
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
61
Figure 2. Response latency for each outcome type. Latencies for each outcome type have been
partially normalized by subtracting the mean response latency. Open bars indicate the condition
with no near wins, and gray bars the condition with near wins. Error bars are standard errors.
for each outcome type for each near-win
condition. In order to assess the effect of
outcome type on response latency, each par-
ticipant’s mean response latency for a given
outcome type was partially normalized by
subtracting their mean response latency for
that condition from it. The means of these
difference scores are presented in Figure 2.
As can be seen in Figure 2, in both con-
ditions, the mean response latency following
wins was longer than that following losses
and in the near-win present condition mean
response latency following near wins was
longer than that following losses but not as
long as that following wins. The direction of
the difference in response latencies between
near-win and clear loss outcomes was very
consistent at the individual level with 91 %
of the participants showing this effect.
Inferential statistics also confirmed this
pattern of results. A paired samples t-test
revealed a significant difference between
mean win and clear loss response latencies
in the near-win-absent condition (t(23) =
8.71, p <0.05, d = 1.91). A repeated
measures ANOVA also revealed a signifi-
cant effect of outcome type on response la-
tencies in the near-win-present condition (F
(2, 46) = 33.36, p <0.05, ηp2 = .59). In addi-
tion, three post-hoc paired samples t-tests
revealed significant differences between win
and clear loss response latencies (t(23) =
7.57, p < 0.05, d = 1.92), win and near win
response latencies (t(23) = 4.67, p < 0.05, d
= 1.46) as well as near win and clear loss
response latencies (t(23) = 3.41, p < 0.05, d
= 0.81) in the near-win-present condition.
These results are consistent with near
wins as conditioned reinforcers and not con-
sistent with near wins as producing frustra-
tion in the current procedure. This result was
consistent with that of Peters et al. (2010)
who found rats produced longer latencies
following near wins than following losses on
a slot machine analog task. In contrast, this
result differs from Dixon and Schreiber
(2004) who found no consistent effect of
near wins on response latencies, and from
7
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
62
SLOT MACHINE NEAR WIN FREQUENCY
Dixon et al. (2013) who found shorter re-
sponse latencies following near wins. These
differences suggest that features of how out-
comes are arranged on slot machines influ-
ence the behavioral effects of near wins.
These features will be discussed further in
the general discussion. Consistent with pre-
vious research Experiment 1 also found that
participants paused longer after experiencing
wins than after experiencing clear losses
(Delfabbro & Winefield, 1999a; Peters et
al., 2010).
The effect of the presence of near wins on
sensitivity to win ratio
The matching law (Baum, 1974) was
used to characterise each subject’s sensitivi-
ty to the relative frequency of wins on each
reel. The matching law refers to the follow-
ing relationship between the distribution of
responses across two alternatives and the
distribution of reinforcers across those two
alternatives:
log(𝐵1
𝐵2) = 𝑎log(𝑅1
𝑅2) + log𝑘 (1)
In the current experiment, B1 was the num-
ber of spins of the last 30 in a given condi-
tion made on the top reel, and B2 the number
of spins of the last 30 in a given condition
made on the bottom reel. R1 was the total
number of wins delivered on the top reel
during a condition and R2 the total number
delivered on the bottom reel. The mean
number of spins made in a condition was 76,
and therefore the last 30 spins represented
39% of each condition on average (range:
29% -52%). If plotted, Equation 1 is the
formula for a straight line, and a is the slope
of that line which also describes how sensi-
tive the distribution of a subject’s behavior
was to the distribution of wins. Occasional-
ly, participants either made no responses on
one of the two reels during a condition or
received no wins from one of the two reels
during a condition. When this occurred, we
added 0.25 to each of R1, R2, B1 and B2 in
order to allow Equation 1 to be used.
We calculated two sensitivity values for
each participant using linear regression: one
for conditions during which near wins were
present, and another for conditions during
which no near wins were present. The mean
sensitivity value was 0.20 (range:-0.93 to
0.73) when near wins were absent and 0.39
when near wins were present (range: -0.15
to 1.27). The average r-squared value was
0.47 (range: 0.01 to 0.99) when near wins
were absent and 0.57 (range: 0.08 to 0.99)
when near wins were present. Figure 3 pre-
sents differences in the individual partici-
pants’ sensitivity values when near wins
were present and their sensitivity values
when near wins were absent. Bars above the
x axis indicate that sensitivity was greater
when near wins were present. As Figure 3
indicates, approximately two thirds of par-
ticipants were more sensitive to the relative
distribution of wins when near wins were
also present in the condition than when they
were absent. A paired samples t-test (t(23) =
2.19, p < 0.05, d = 0.45) confirmed that sen-
sitivity values were significantly greater dur-
ing the condition in which near wins were
present.
The majority of sensitivity values great-
er than zero demonstrate the sensitivity of
participants to the ratio of wins presented on
a slot machine analog task. This is consistent
with the findings of Lie et al. (2009) who
found that humans were sensitive to the rate
of reinforcers in a similar rapidly-changing
choice paradigm in a non-gambling context.
The current experiment and the results of
Lie et al. confirm the utility of this proce-
dure for efficiently assessing humans’ sensi-
tivity to changing reinforcement rate in a
given context, extending the use of this pro-
cedure, originated by Davison and Baum
(2000), to a context of applied relevance.
As displayed in Figure 1 the top slot on
each version of the slot machine was
8
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
63
Figure 3. Differences between sensitivity to relative win frequency when near wins were present
and sensitivity to relative win frequency when near wins were absent for each participant. Bars
above the x axis indicate participants exhibited higher sensitivity to win rate ratios when near-
wins were also present in the condition, those below the axis indicate participants were more
sensitive to relative win frequency when near wins were absent.
always fruit symbols and the bottom slot
was always Viking symbols. The matching
law analysis allowed an assessment of
whether participants showed a bias towards
one or other of these reels. A bias is a pref-
erence for one of the response alternatives
(here, responding on one of the two reels)
that is unrelated to the rate of reinforcement
(wins) presented by those two alternatives
(Baum, 1974). There was no consistent
across-participant pattern of biases to one or
other of the reels, suggesting that neither the
position (top or bottom) of a reel nor the
symbol set presented on that reel consistent-
ly affected participants’ preference for that
reel.
EXPERIMENT 2
Experiment 2 investigated whether re-
sponse latency and sensitivity to wins were
affected by changes in the frequency of
near-win outcomes. Kassinove and Schare
(2001) found that increases in the proportion
of near wins initially increased but later de-
creased persistence of play. Decreases may
therefore also occur in sensitivity to the rela-
tive frequency of wins or in response latency
length when the proportion of near wins ex-
perienced is above a particular value. To in-
vestigate this possibility we conducted a
second experiment identical to Experiment
1, except that players experienced one ses-
sion where near wins were present on 25%
of non-win trials and another where they
were present on 50% of non-win trials.
These values were selected because previous
9
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
64
SLOT MACHINE NEAR WIN FREQUENCY
research indicates that persistent play effects
are greatest when the near win frequency
lies somewhere between 25 and 50% (Chan-
tal, Vallerland, Ladouceur & Ferland, 1996;
Côte et al., 2003; Kassinove & Schare,
2001).
METHOD
Participants
Twenty-four first year psychology stu-
dents from Victoria University of Welling-
ton completed Experiment 2 in partial ful-
fillment of a course requirement.
Apparatus
The materials used were as for Experi-
ment 1.
Procedure
The procedure was as for Experiment 1,
except participants completed one session
during which near wins occurred on 25% of
non-win trials and another where near wins
occurred on 50% of non-win trials. The or-
der in which participants experienced these
two conditions was counterbalanced. Of the
192 conditions played in Experiment 2, 159
ended due to the acquisition of 12 wins and
32 conditions ended after reaching the 10-
minute time limit for the condition. Data
from one of the 192 conditions were lost due
to a recording error.
RESULTS AND DISCUSSION
Mean response latencies were calculat-
ed as for Experiment 1. The within-
condition pattern of mean response latencies
found in Experiment 1 was replicated in Ex-
periment 2 with response latencies for near
wins falling between those for wins and
losses in both conditions (see Figure 4). A
clear majority of participants showed this
difference in response latency in each condi-
tion, and there was no difference in the dis-
tribution of response latencies between con-
ditions. A 2 (25% near wins, 50% near wins)
x3 (clear loss, near win, win) repeated
measures ANOVA confirmed that there was
no significant interaction of near win pro-
portion by outcome type (F (2, 46) =1.69, p
= 0.20, ηp2 = 0.068) and no significant main
effect of near win proportion on response
latencies (F (1, 23) = 0.58, p = 0.46, ηp2 =
0.025). The 2x3 ANOVA did however re-
veal a significant main effect of outcome
type on response latencies (F (2, 46) =
27.70, p < 0.05, ηp2 = 0.46). Following this,
post-hoc t-tests revealed significant differ-
ences between mean response latencies of
wins and clear losses (t(23) = 8.28, p < 0.05,
d = 1.31), wins and near wins (t(23) = 5.00,
p < 0.05, d = 0.94), as well as near wins and
clear losses (t(23) = 3.15, p < 0.05, d =
0.52).
The pauses following win, near win and
clear loss outcomes in Experiment 2 repli-
cate the pattern of results found in Experi-
ment 1, extending this finding to an addi-
tional near win frequency (25%). Pause
length was not affected by the relative pro-
portion of near wins experienced. The in-
consistent effects of near wins on pause
length in Dixon and Schreiber (2004) is
therefore unlikely to be due to differences in
the proportions of near wins experienced by
each participant.
As in Experiment 1, sensitivity values
were calculated for each participant in each
near win frequency condition. In Experiment
2, the mean number of spins made in a con-
dition was 71, and therefore the final 30
spins that were included in calculations of
sensitivity represented 42% (range: 29% to
66%) of the condition on average. The mean
sensitivity value was 0.09 (range:-0.34 to
0.64) when near wins were presented on
25% of trials and 0.20 when near wins were
presented on 50% of trials (range: -0.17 to
0.71). For two participants, r-squared could
not be calculated. For the remaining partici-
pants, the average r-squared value was 0.41
10
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
65
Figure 4. Response latency for each outcome type. Latencies for each outcome type have been
partially normalized by subtracting the mean response latency. Open bars indicate the condition
with 25% near wins, and gray bars the condition with 50% near wins. Error bars are standard er-
rors.
(range: 0 to 0.95) when near wins were ab-
sent and 0.44 (range: 0 to 0.9) when near
wins were present. Figure 5 displays the dif-
ferences in the individual participants’ sensi-
tivity values when near wins were present
on 50% of trials and their sensitivity values
when near wins were present on 25% of tri-
als. Bars above the x axis indicate that sensi-
tivity was greater when near wins were pre-
sent on 50% of trials. The majority of bars
on Figure 5 are above zero indicating that
most participants were more sensitive to the
relative distribution of wins when near wins
were presented on 50% rather than 25% of
non-winning trials. A paired samples t-test (t
(23) = 2.484, p < .05, d = 0.51) confirmed
that sensitivity values were significantly
greater in the 50% near win condition. This
finding extends the results of Experiment 1
by indicating that incremental increases in
near win frequency produce incremental in-
creases in sensitivity to win frequency. As in
Experiment 1, there was no consistent bias
for either symbol set.
GENERAL DISCUSSION
The current study found that partici-
pants produced longer response latencies
following near wins than following clear
losses, an effect previously observed by Pe-
ters et. al. (2010) with rats but not previous-
ly observed with humans. The current study
11
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
66 SLOT MACHINE NEAR WIN FREQUENCY
Figure 5. Differences between sensitivity to relative win frequency when near wins were present
on 50% of trials and sensitivity to relative win frequency when near wins were present on 25%
of trials for each participant. Bars above the x axis indicate higher sensitivity to win rate ratios
when near-wins were present on 50% of trials, those below indicate participants were more sen-
sitive to relative win frequency when near wins were present on 25% of trials.
also found that near wins increased sensitivi-
ty to rate of wins. These results are con-
sistent with near wins acting as conditioned
reinforcers rather than producing frustration
in the current arrangement. If the near wins
had produced frustration (Amsel, 1958) then
pauses following them would have been
shorter than those following clear losses,
and no systematic effect on sensitivity to
reinforcement ratios would have been ex-
pected.
The longer latencies observed in the
current study differed from the results of
Dixon and Schreiber (2004) who found no
consistent pattern of response latencies, and
from that of Dixon et al. (2013) who found
shorter latencies following near wins than
other losses. Together, these studies suggest
that the behavioral effects of near wins de-
pend on features of the slot machine pro-
gram, and the outcomes and related symbols
presented. In the current study, wins and
near wins were both presented during play
and near wins appeared to function as condi-
tioned reinforcers. In Dixon et al.’s proce-
dures near wins were presented without the
wins they resembled and they appeared to
elicit frustration. This may suggest that in
the presence of wins, near wins develop
conditioned reinforcement effects but in the
absence of wins, near wins produce frustra-
tion. Future research systematically manipu-
lating the frequency of wins and near wins
could clarify this. Ghezzi et al. (2006) inves-
tigated the effects of multiple combinations
of win size and near-win frequency on per-
sistence of gambling. Results were incon-
12
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
67
sistent, underscoring the complexity of the
issue.
Additionally, procedures in which near
wins resemble wins for longer portions of
outcome sequences may be more likely to
establish those wins as conditioned reinforc-
ers. In the current, five-symbol slot machine
analog in which only near wins with four of
five symbols matching were included, near
wins resembled wins for larger portions of
outcome sequences than they had in previ-
ous arrangements. The mixed results of
Dixon and Schreiber (2004) may have re-
flected the fact that they did not separate out
trial types on which the pattern of symbols
presented differed in meaningful ways.
There is, however, a possible alternative
explanation to conditioned reinforcement for
the differential pauses we observed. The
longer pauses following near wins might
simply be an artefact of the sequential
presentation of symbols in the outcome
stimuli in combination with the fact that par-
ticipants require some processing time be-
fore selecting their next bet amount and al-
ternative. This processing time may begin
when the outcome of the previous spin is
known rather than when the opportunity to
make the next spin is presented. If this is the
case, then, following clear loss outcomes,
this processing time may begin earlier, while
the remaining symbols are displayed and
thus produce apparently shorter pauses fol-
lowing these outcomes than near wins. This
explanation, however, does not account for
the difference in pause length between wins
and near wins as both types of outcomes are
revealed when the last symbol is displayed.
Nevertheless this possible explanation re-
mains and could be evaluated by replicating
this study with simultaneous presentation of
all symbols.
This study also found that higher rates
of near wins produced increased sensitivity
to the relative frequency of wins. Previous
research suggests two possible explanations
for this. Firstly, conditioned reinforcement
may explain this effect as it does for the in-
creased pauses. Alsop and Elliffe (1988)
found that when pigeons were responding on
concurrent VI VI schedules increasing the
overall reinforcement rate while keeping the
reinforcement rate ratios equal produced
higher sensitivity values. This result sug-
gests that increasing overall reinforcement
rate in a gambling context may increase sen-
sitivity. If near wins are conditioned rein-
forcers, then conditions in which they oc-
curred more frequently had higher overall
effective reinforcement rates, and, therefore
perhaps, higher sensitivity. This conclusion
is tentative given the difference in the pro-
cedure through which reinforcement rate
was increased across the two studies (the
current procedure added equal rates of near
wins to both alternatives).
An alternative possibility is suggested
by an experiment conducted by Madden and
Perone (1999). In that study, requiring par-
ticipants to attend to schedule-correlated
stimuli increased sensitivity to reinforce-
ment rate ratios. The addition of near wins
may have had a similar effect because it led
participants to increase their attentiveness to
the gambling outcome stimuli in order to
discriminate wins from physically-similar
near-wins.
Although there are alternative explana-
tions to conditioned reinforcement for the
effects of near wins on both response laten-
cy and sensitivity to wins in the current
study, conditioned reinforcement as an ex-
planation has the advantage of parsimony in
that it alone accounts for both response la-
tency and reinforcement sensitivity effects.
Future research could investigate the extent
to which stimulus generalization or condi-
tioning that occurs within a winning spin
contribute by systematically varying the ex-
tent to which near wins are paired with, ver-
sus physically similar to, wins. The extent to
which each of these processes contributes to
13
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
68
SLOT MACHINE NEAR WIN FREQUENCY
the near win effect has implications for un-
derstanding the importance of this effect for
problem gamblers. If pairing is crucial, then
the effect may be stronger in problem gam-
blers as they have experienced many win
outcomes and therefore many pairings of
near wins and wins.
An important novel feature of the cur-
rent procedure was the application of a rap-
idly- changing choice procedure in combina-
tion with the generalized matching law to
assess sensitivity to wins. Sensitivity values
were between zero and one (undermatching)
consistent with previous findings with
humans but closer to indifference than
those found by Lie, et al. (2009). Here, the
strongest mean sensitivity of any condition
was 0.38 (in the 50% near wins condition in
Experiment 1), while the mean sensitivity
they observed was 0.52. This may reflect the
fact that behavior in the gambling context is
uniquely influenced by factors other than
reinforcement distribution such as inaccu-
rate, self-generated verbal rules. Condi-
tioned reinforcement and verbal rules may
interact in determining the effect of near
wins on gambling. Research (e.g. Dymond
& Roche, 2010; Dymond, McCann, Grif-
fiths, Cox & Crocker, 2012) has shown that
derived verbal relations can influence gam-
bling behavior. Directly relevant to near
wins, Dixon, Nastally, Jackson and Habib
(2009) found that participants who acquired
a derived relation between an image of a
near win and the word “almost” rated nears
wins as more “win like” than they had be-
fore they underwent relational training. If
gamblers have acquired the (inaccurate)
verbal rule that near wins indicate that addi-
tional gambling is more likely to produce a
win, then near wins might spur persistent
play. Future research could identify experi-
ences that lead near wins to increase the per-
sistence of gambling through either or both
of these processes.
The current findings suggest that near
wins are conditioned reinforcers because
they both produced longer pauses than clear
losses and increased sensitivity to win fre-
quency. The increased reinforcement rate
created by slot machine operators’ addition
of near wins is therefore likely the mecha-
nism through which near wins increase the
persistence of slot machine play. Future re-
search that further investigates this process
will contribute to the design of regulations
and interventions to reduce the adverse so-
cial impact of slot machines by reducing
persistence.
REFERENCES
Alsop, B., & Elliffe, D. (1988). Concurrent
schedule performance: Effects of rela-
tive and overall reinforcer rate. Journal
of the Experimental Analysis of Behav-
ior, 49, 21-36.
Amsel, A. (1958). The role of frustra-
tivenonreward in noncontinuous reward
situations. Psychological Bulletin, 55,
102-119.
Baum, W. M. (1974). On two types of de-
viation from the matching law: Bias and
undermatching. Journal of the Experi-
mental Analysis of Behavior, 22, 231-
242.
Breen, R. B., & Zimmerman, M. (2002).
Rapid onset of pathological gambling in
machine gamblers. Journal of Gam-
bling Studies, 18, 31-43.
Chantal, Y., Vallerland, R., Ladouceur, R.,
& Ferland, F. (1996). How close is
close enough? Near misses perceptions
and gambling. Paper presented at the
104th Annual Convention of the Amer-
ican Psychological Association, Toron-
to, Ontario, Canada.
Chóliz, M. (2010). Experimental analysis of
the game in pathological gamblers: Ef-
fect of the immediacy of the reward in
slot machines. Journal of Gambling
Studies, 26, 249-256.
14
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
TADHG E. DALY ET AL.
69
Côte, D., Caron, A., Aubert, J., Desrochers,
V., & Ladouceur, R. (2003). Near wins
prolong gambling on a video lottery
terminal. Journal of Gambling Studies,
19, 433-438.
Davison, M. & Baum, W.M. (2000) Choice
in a variable environment: every rein-
forcer counts. Journal of the Experi-
mental Analysis of Behavior, 74, 1-24.
Delfabbro, P. H., & Winefield, A. H.
(1999a). Poker-machine gambling: An
analysis of within session characteris-
tics. British Journal of Psychology, 90,
425-439.
Delfabbro, P. H., & Winefield, A. H.
(1999b). The danger of over-
explanation in psychological research:
A reply to Griffths. British Journal of
Psychology, 90, 447-450.
Dixon, M. J., MacLaren, V., Jarick, M.,
Fugelsang, J. A. & Harrigan, K. A.
(2013). The Frustrating Effects of Just
Missing the Jackpot: Slot Machine
Near-Misses Trigger Large Skin Con-
ductance Responses, But No Post-
reinforcement Pauses. Journal of Gam-
bling Studies, 29, 661-674.
Dixon, M. J., Harrigan, K. A., Jarick , M.,
MacLaren, V., Fugelsang, J. A., &
Sheepy, E. (2011). Psychophysiological
arousal signatures of near-misses in slot
machine play. International Gambling
Studies, 11, 393-407.
Dixon, M. R., Nastally, B. L., Jackson, J. E.,
& Habib, R. (2009). Altering the near-
miss effect in slot machine gamblers.
Journal of Applied Behavior Analysis,
42, 913-918.
Dixon, M. R., & Schreiber, J. E. (2004).
Near-miss effects on response latencies
and win estimations of slot machine
players. The Psychological Record, 54,
335-348.
Dixon, M. R., MacLin, O. H., & Daugherty,
D. (2006). An evaluation of response
allocations to concurrently available
slot machine simulations. Behavior Re-
search Methods, 38, 232-236.
Dymond, S. & Roche, B. (2010). The impact
of derived relational responding on
gambling behavior. Analysis of Gam-
bling Behavior, 4, 38-53.
Dymond, S., McCann, K., Griffiths, J., Cox,
A., & Crocker, V. (2012). Emergent re-
sponse allocation and outcome ratings
in slot machine gambling. Psychology
of Addictive Behaviors, 26, 99-111.
Ferris, J. & Wynne, H. (2001). The Canadi-
an Problem Gambling Index: Final Re-
port. Canadian Centre on Substance
Abuse.
Ferster, C. B., & Skinner, B. F. (1957).
Schedules of Reinforcement. Eng-
lewood Cliffs: Prentice-Hall.
Ghezzi, P. M., Wilson, G., R., Porter, C. K.
(2006). The near-miss effect in simulat-
ed slot machine play. In P.M. Ghezzi,
C.A. Lyons, M.R. Dixon, & G.R. Wil-
son (eds.) Gambling: Behavior Theory,
Research, and application (pp.155-
170). Context Press: Reno, NV.
Griffiths, M. D. (1999). The psychology of
the near-miss (revisited): A comment
on Delfabbro & Winefield (1999). Brit-
ish Journal of Psychology, 90, 441-445.
Harrigan, K. A. (2007). Slot machine struc-
tural characteristics: Distorted player
views of payback percentages. Journal
of Gambling issues, 20, 215-234.
Harrigan, K. A. (2008). Slot machine struc-
tural characteristics: Creating near
misses using high award symbol ratios.
International Journal of Mental Health
and Addiction, 6, 353-368.
Haw, J. (2008). The relationship between
reinforcement and gaming machine
choice. Journal of Gambling Studies,
24, 55-61.
15
Daly et al.: Slot Machine Near Wins: Effects on Pause and Sensitivity to Win R
Published by theRepository at St. Cloud State, 2014
70
SLOT MACHINE NEAR WIN FREQUENCY
Jenkins, W. O., & Stanley, J. C. (1950). Par-
tial reinforcement: A review and cri-
tique. Psychological Bulletin, 47, 193-
234.
Kahneman, D., & Tversky, A. (1982). Psy-
chology of preferences. Scientific
American, 246, 160-173.
Kassinove, J. I., & Schare, M. L. (2001).
Effects of the “near miss” and the “big
win” on persistence at slot machine
gambling. Addictive Behavior, 15, 155-
158.
Ladouceur, R. & Giroux, I. (2006). The ef-
fect of near wins on the choice of a vid-
eo lottery terminal. Gambling Research,
18, 69-75.
Ladouceur, R. & Sevigny, I. (2002). Sym-
bols presentation modality as a deter-
minant of gambling behavior. Journal
of Psychology: Interdisciplinary and
Applied, 136, 443-448.
Lie, C. Harper, D. N., Hunt, M. (2009).
Human performance on a two-
alternative rapid-acquisition choice
task. Behavioural Processes, 81, 244–
249
Loftus, G. R, & Loftus, E. F. (1983). Mind
at Play: The Psychology of Video
Games. New York: Basic Books.
Madden, G. J., & Perone, M. (1999). Human
sensitivity to concurrent schedules of
reinforcement: effects of observing
schedule-correlated stimuli. Journal of
the Experimental Analysis of Behavior,
71, 303-318.
Peters, H., Hunt, M., Harper, D. (2010). An
animal model of slot machine gam-
bling: The effect of structural character-
istics on response latency and persis-
tence. Journal of Gambling Studies, 26,
521-531.
Reid, R. L. (1986). The psychology of the
near miss. Journal of Gambling Behav-
ior, 21, 32-39.
Skinner, B. F. (1953). Science and human
behavior. New York: Macmillan.
Wood, R.T.A., & Griffiths, M.D. (2007). A
qualitative investigation of problem
gambling as an escape-based coping
strategy. Psychology and Psychothera-
py: Theory, Research and Practice, 80,
107-125. doi:
10.1348/147608306X107881
Yu, C.Y. (2002). Evaluating cutoff criteria
of model fit indices for latent variable
models with binary and continuous out-
comes. Dissertation for the degree
Doctor of Philosophy in Education.
University of California, Los Angeles.
Action Editor: Jeffrey N. Weatherly
16
Analysis of Gambling Behavior, Vol. 8 [2014], Art. 1
https://repository.stcloudstate.edu/agb/vol8/iss2/1
... Either way, they could serve as motivators that prolong gamblers' behaviors, despite not winning. When the effects of near-miss outcomes were investigated on a computer-simulated slot machine, on which the win frequency was also manipulated, latencies were found to be longer following near misses (Daly et al., 2014). Latency tends to follow reinforced trials, but not unreinforced. ...
... Two other phenomena that complicate the consequential side of learning are near misses and losses disguised as wins. These are events where an omission of what is expected to be a reinforcer actually functions as a reinforcer for gambling behavior (Barton et al., 2017;Daly et al., 2014;Foxall & Sigurdsson, 2012). This is a phenomenon that is also highly relevant not only for understanding gambling behavior, but for understanding learning processes in themselves, since it appears to contradict the keystones of learning theory. ...
Full-text available
Article
Gambling is a field that harbors both harmless recreational activities and pathological varieties that may be considered an addictive disorder. It is also a field that deserves special interest from a learning theoretical perspective, since pathological gambling represents both a pure behavioral addiction involving no ingestion of substances and behavior that exhibits extreme resistance to extinction. As the field of applied psychology of learning, or behavior analysis, espouses a bottom-up approach, the basis of understanding begins in basic research on behavioral principles. This article provides a narrative review of the field of laboratory experiments conducted to disentangle the learning processes of gambling behavior. The purpose of this review is to give an overview of learning principles in gambling that has been demonstrated under lab conditions and that may be of importance in the development of clinical applications when gambling has become a problem. Several processes, like the importance of delay and probability discounting, reinforcement without actual winning, and rule governed behavior have been experimentally verified. The common denominator appears to be that they impede extinction. Other areas, especially Pavlovian conditioning, are scarce in the literature. Our recommendations for the future would be to study Pavlovian and instrumental conditioning in interaction. Treatment programs should profit from strategies that serve to enhance extinction learning. We also conclude that online gambling should provide a promising environment for controlled research on how to limit excessive gambling, provided that the gambling companies are interested in that.
... A significant research effort has focused on how contextual stimuli drive preferences in equivalent concurrent slot machines (Nastally, Dixon, & Jackson, 2010; Zlomke & Dixon, 2006). Others have focused on the effect of different types of stimulus, such as near misses (Daly et al., 2014; Ghezzi et al., 2006; Reid, 1986) (van Holst, Chase, & Clark, 2014), big wins (Kassinove & Schare, 2001), losses disguised as wins (Dixon, Harrigan, Sandhu, Collins, & Fugelsang, 2010), or the structural features of gambling games (Griffiths & Auer, 2013) and their effect on behaviour. Many of these studies have looked at different aspects of gambling, such as machine preference (Dymond, McCann, Griffiths, Cox, & Crocker, 2012), rate of gambling (Dixon et al., 2010), post reinforcement pauses (Delfabbro & Winefield, 1999), latencies between gambles (James, O'Malley, & Tunney, in press), fixed interval schedules in betting (Dickerson, 1979), the random ratio schedule of reinforcement (Crossman, Bonem, & Phelps, 1987; Haw, 2008; Hurlburt, Knapp, & Knowles, 1980) and perseverance during extinction (). ...
... The other alternative is near-misses, where a similar component to drug seeking has been proposed (Ghezzi et al., 2006). It has been alternatively proposed that near-misses get their predictive value from winning outcomes i.e. near-misses on a slot machine must occur prior to a win (Daly et al., 2014 ), largely in the same manner as arousal. Additionally the two interact ; studies have shown greater levels of autonomic arousal in recreational gamblers to losses disguised as wins (Dixon et al., 2010), and greater reactivity to near-misses in problem gamblers (Dymond et al., 2014; van Holst et al., 2014). ...
Full-text available
Article
This manuscript overviews the behavioural (i.e. associative learning, conditioning) research in behavioural addictions, with reference to contemporary models of substance addiction and ongoing controversies in the behavioural addictions literature. The role of behaviour has been well explored in substance addictions and gambling but this focus is often absent in other candidate behavioural addictions. In contrast, the standard approach to behavioural addictions has been to look at individual differences, psychopathologies and biases, often translating from pathological gambling indicators. An associative model presently captures the core elements of behavioural addiction included in the DSM (gambling) and identified for further consideration (internet gaming). Importantly, gambling has a schedule of reinforcement that shows similarities and differences from other addictions. While this is more likely than not applicable to internet gaming, it is less clear whether it is so for a number of candidate behavioural addictions. Adopting an associative perspective, this paper translates from gambling to video gaming, in light of the existing debates on this matter and the nature of the distinction between these behaviours. Finally, a framework for applying an associative model to behavioural addictions is outlined, and it's application toward treatment.
... One example of this is the matching law (Herrnstein, 1974) and its generalization (Baum, 1974), which attempts to describe how organisms distribute responding to multiple concurrent ratio or interval schedules. There is a literature on response allocation in concurrent slot machines, but findings in this area have been mixed; a number of studies (Coates & Blaszczynski, 2014;Daly et al., 2014;Dixon, Fugelsang, MacLaren, & Harrigan, 2013;Dixon, MacLin, & Daugherty, 2006;Dymond, McCann, Griffiths, Cox, & Crocker, 2012;Zlomke & Dixon, 2006) found evidence consistent with matching, but there is also evidence gamblers undermatch, showing greater (or in some cases, total) equivalence between machines that diverge either in rate of return to player or rate of reinforcement on a ratio schedule (Coates & Blaszczynski, 2013;Daly et al., 2014;Lucas & Singh, 2012;Weatherly, Thompson, Hodny, Meier, & Dixon, 2009). In addition, matching is highly susceptible to being overridden by contextual cues (Nastally, Dixon, & Jackson, 2010;Zlomke & Dixon, 2006) although this appears to weaken with extended exposure to the contingencies of a machine (Hoon & Dymond, 2013). ...
... One example of this is the matching law (Herrnstein, 1974) and its generalization (Baum, 1974), which attempts to describe how organisms distribute responding to multiple concurrent ratio or interval schedules. There is a literature on response allocation in concurrent slot machines, but findings in this area have been mixed; a number of studies (Coates & Blaszczynski, 2014;Daly et al., 2014;Dixon, Fugelsang, MacLaren, & Harrigan, 2013;Dixon, MacLin, & Daugherty, 2006;Dymond, McCann, Griffiths, Cox, & Crocker, 2012;Zlomke & Dixon, 2006) found evidence consistent with matching, but there is also evidence gamblers undermatch, showing greater (or in some cases, total) equivalence between machines that diverge either in rate of return to player or rate of reinforcement on a ratio schedule (Coates & Blaszczynski, 2013;Daly et al., 2014;Lucas & Singh, 2012;Weatherly, Thompson, Hodny, Meier, & Dixon, 2009). In addition, matching is highly susceptible to being overridden by contextual cues (Nastally, Dixon, & Jackson, 2010;Zlomke & Dixon, 2006) although this appears to weaken with extended exposure to the contingencies of a machine (Hoon & Dymond, 2013). ...
Full-text available
Article
This manuscript reviews the extant literature on key issues related to mobile gambling and considers whether the potential risks of harm emerging from this platform are driven by pre-existing comorbidities or by psychological processes unique to mobile gambling. We propose an account based on associative learning that suggests this form of gambling is likely to show distinctive features compared with other gambling technologies. Smartphones are a rapidly growing platform on which individuals can gamble using specifically designed applications, adapted websites or text messaging. This review considers how mobile phone use interacts with psychological processes relevant to gambling, the games users are likely to play on smartphones, and the interactions afforded by smartphones. Our interpretation of the evidence is that the schedules of reinforcement found in gambling interact with the ways in which people tend to use smartphones that may expedite the acquisition of maladaptive learned behaviours such as problem gambling. This account is consistent with existing theories and frameworks of problem gambling and has relevance to other forms of mobile phone use.
... J. Dixon et al., 2011Dixon et al., , 2013Sharman & Clark, 2016;Stange et al., 2016Stange et al., , 2017. Previous work examining how quickly gamblers initiate a new gamble after a 'near miss' (an indicator of response vigor) have also yielded inconsistent results (Belisle & Dixon, 2016;Daly et al., 2014;M. J. Dixon et al., 2013;Stange et al., 2016Stange et al., , 2017Worhunsky et al., 2014). ...
Full-text available
Article
Both humans and some non-human animals tend to respond more vigorously after failing to obtain rewards. Such response invigoration becomes more pronounced when individuals have increased expectations of obtaining rewards during reward pursuit (expectancy), and when they perceive the eventual loss to be proximal to reward receipt (proximity). However, it was unclear whether proximity and expectancy may have distinct influences on response vigor. To investigate this question, we developed a computerized ’scratch card’ task, in which participants turned three cards one by one and won points when all three cards matched (AAA). After each game, they pressed keys to confirm the outcome and start a new game. We included three types of losses: AAB, where participants had increased expectancy of winning as the game evolved, and the final outcome was proximal to winning; ABB and ABA, with reduced expectancy, but high proximity to winning; and ABC, with reduced expectancy and low proximity to winning. In three online studies, we consistently observed that participants confirmed losses more quickly than wins. Importantly, detailed analyses of the different types of losses revealed that proximity reduced vigor, whereas expectancy increased it. Together, these findings are in line with general appraisal theories: the adjustments of response vigor may be triggered by the appraised discrepancy between the current state and a reference state (e.g., attaining one’s goal), and serve to close the gap and facilitate goal pursuit. These findings may also have implications for the effect of ‘near miss’ on gambling persistence. Further exploring how reward omission impacts response vigor may help us better understand the goal pursuit process, and how it becomes maladaptive under certain circumstances.
... Furthermore, near misses can be conceptualized globally (i.e., cherry-cherry-lemon could be viewed as a single stimulus) or locally (i.e., each element as a separate stimulus). From a global view, the conditionally reinforcing effect of win-related stimuli may generalize better to near misses than to other, more dissimilar, misses (Belisle and Dixon 2016;Daly et al. 2014). Evidence consistent with a stimulus generalization account comes from findings that reel outcomes more visually similar to wins will generate longer latencies that are more like the latencies which occur after a win. ...
Full-text available
Article
In games of chance, a near miss is said to occur when feedback for a loss approximates a win. For instance, obtaining "cherry-cherry-lemon" on a slot machine could be considered a near miss. Sixty-six years ago, B.F. Skinner first proposed the idea that near-miss events might reinforce continued play in slot machines, and despite some inconsistencies in the experimental literature, belief in this "near-miss effect" has remained strong. In the present manuscript, we will review this literature and present experimental assessments of the near-miss effect on the frequency of the gambling response. Experiment 1 used a tightly controlled resistance-to-extinction procedure in pigeons to evaluate the putative reinforcing effect of near misses relative to a control "far-miss" reel pattern. Experiment 2 extended Experiment 1's procedure to human participants. The results of both experiments failed to support the near-miss effect hypothesis. Experiment 3 used a further simplified procedure to assess the validity of the resistance-to-extinction paradigm when a probable conditional reinforcer was present on the reel stimuli. Although a clear conditional response was obtained from the reel, subsequent testing in extinction revealed no conditionally reinforcing function of this stimulus on operant response frequency.
... THE NEAR-MISS EFFECT IN SLOT MACHINES 7 Furthermore, near misses can be conceptualized globally (e.g., cherry-cherry-lemon could be viewed as a single stimulus) or locally (each element as a separate stimulus). From a global view, the conditionally reinforcing effect of win-related stimuli may generalize better to near misses than to other, more dissimilar, misses (Belisle & Dixon, 2016;Daly et al., 2014). ...
Preprint
In games of chance, a near miss is said to occur when feedback for what is otherwise a loss approximates a win. For instance, obtaining “cherry-cherry-lemon” on a slot machine could be considered a near miss. Sixty-six years after B. F. Skinner first proposed the idea that near-miss events might be reinforcing play in slot machines, belief in this ‘near-miss effect’ has remained strong despite the troublesome experimental literature. Rather than inferring its effects on behaviour, the present study reviewed and experimentally assessed the near-miss effect as it pertains to the gambling response. Experiment 1 used a tightly controlled resistance-to-extinction procedure in pigeons to evaluate the putative reinforcing effect of near misses relative to a control “far-miss” reel pattern. Experiment 2 extended Experiment 1’s procedure to human participants. The results of both failed to support the near-miss effect hypothesis. Experiment 3 used a further simplified procedure to assess the validity of the resistance-to-extinction paradigm when a probable conditional reinforcer was present on the reel stimuli. Although a clear discriminative function was obtained from the reel, subsequent testing in extinction revealed no reinforcing function of this stimulus.
... Studies of the postreinforcement pause have identified mixed findings. Some have found no or limited evidence for an effect [18,19] , whereas others have found that near-misses are perceived intermediately between wins and losses [20,41] . The findings of this study suggest evidence for the latter, despite findings on scratchcard play thus far have been equivocal [19,20] . ...
Full-text available
Article
Smartphone users engage extensively with their devices, on an intermittent basis for short periods of time. These patterns of behaviour have the potential to make mobile gambling especially perseverative. This paper reports the first empirical study of mobile gambling in which a simulated gambling app was used to measure gambling behaviour in phases of acquisition and extinction. We found that participants showed considerable perseverance in the face of continued losses that were linearly related to their prior engagement with the app. Latencies between gambles were associated with the magnitude of reinforcement; more positive outcomes were associated with longer breaks between play and a greater propensity to end a gambling session. Greater latencies were associated with measurements of problem gambling, and perseverance with gambling-related cogni-tions and sensation-seeking behaviour.
... A near-miss is an oft-studied phenomenon in which losses in gambling that are topographically closer to wins take on subjective properties of wins (e.g., two out of three matching icons on a three-reel slot machine; a good hand in poker that loses on the last card, etc.). Near-misses have often been conceptualized as conditioned reinforcers that may maintain gambling behavior (Daly, Tan, Hely, Macaskill, Harper, & Hunt, 2015; Dymond et al., 2014; Foxall & Sigurdsson, 2012; Habib & Dixon, 2010; cf., Witts, Ghezzi, & Manson, 2015). The modification of the disordered gamblers' near-miss ratings may have contributed to curbing their gambling behavior. ...
Article
Acceptance and Commitment Therapy for Pathological Gamblers (Dixon & Wil-son, 2014) is a treatment manual for applying acceptance and commitment therapy (ACT) to disordered gambling. Utility of the manual will be gauged over time, as empirical data on the evaluation of ACT for disordered gambling were not available at the time of publication or at the time of review. The manual, while light on theory , allows for a behavioral and ACT-consistent conceptualization of disordered gambling, although such an analysis is not explicit. The present manuscript provides a brief review and reaction to the content with a supplemental conceptualiza-tion of disordered gambling in terms of escape, negative reinforcement, and experi-ential avoidance.
Article
Administration of (±) 3,4-methylenedioxymethamphetamine (MDMA) causes memory errors by increasing proactive interference. This might occur because MDMA alters sensitivity to reinforcement. The current 2 experiments investigated this directly by assessing the acute (Experiment 1) and chronic (Experiment 2) effects of MDMA on sensitivity to reinforcement. We presented 5 pairs of concurrent variable interval schedules within each session and calculated sensitivity to reinforcement for 3 acute doses of MDMA. In contrast to the related drug, d-amphetamine, and in spite of producing reductions in response rate, MDMA did not reduce sensitivity to reinforcement rate. Chronic administration of a fixed dose of MDMA following each session reduced response rate but did not affect sensitivity to reinforcement rate. In combination with previous research, these results indicate that related drugs may have different effects on sensitivity to reinforcement and that these effects should be considered when interpreting disruptions to operant task performance caused by drug administration. (PsycINFO Database Record
Full-text available
Article
Previous research suggests that the actions of a confederate can alter participants' gambling behavior. In the present experiment, male participants played Blackjack either alone or in the presence of a confederate. The confederate either quit early in the session or played for the entire session. Across sessions in which the confederate played for the entire session, how much the confederate bet per hand and how accurately he played were manipulated. Participants gambled significantly more money across the session when a confederate played the entire session than when a confederate left early in the session. During sessions in which the confederate played for the entire session, the number of hands participants played and their total amount bet varied directly with the confederate's bet size. Overall, the results indicate that gambling behavior can be indirectly influenced by another's gambling behavior, which may provide useful data for researchers and therapists seeking to predict and control gambling behavior.
Full-text available
Article
This study investigated whether risk taking on a laboratory gambling task differed depending on whether participants gambled with earned or experimenter-provided game credits. Participants made repeated choices between two options, one to wager game credits on a game that produced probabilistic gains and losses, and one to gain game credits with certainty. Choice was investigated across stake and no-stake conditions and condition order was counterbalanced across conditions. Risk taking was higher under stake than no-stake conditions, but only when stake conditions were experienced first. There was no effect on risk taking of the amount of the certain gain. Results are consistent with previous research showing that participant-stake procedures promote greater risk taking than procedures that allow participants to gamble with their own earnings, and also show that experience gambling with earned credits has an enduring effect on risk taking.
Full-text available
Article
This paper presents a sample three-reel three-coin slot machine game with a bonus for three coins, and a true payback percentage of 85.6% when one or two coins are wagered and 92.5% when three coins are wagered. The player sees the winning or losing combination of three symbols on the payline as well as (a) the physical reels as they scroll by and (b) what is just above and just below the payline at the end of play. An analysis of this game shows that observing the physical reels and what is just above and just below the payline indicates that the slot machine would lose money, and thus the player would make money, as the game would have a payback percentage in the range of 192%-486% if this reflected reality. The paper concludes by discussing the results of the analysis in terms of gaming regulations and problem gambling.
Article
The present study examined the ability to experimentally manipulate the cognitive heuristic entitled the "illusion of control." Five adult human females gambled at roulette for the opportunity to earn course extra-credit points. An alternating treatments design was utilized whereby in one condition subjects were allowed to select the location on the roulette board they placed their bets, and in the other condition subjects had to give their chips to the experimenter to select the location. In addition, after subjects had played the game for a number of trials, inaccurate rules related to this "illusion of control" were introduced in a multiple baseline fashion across subjects, later followed by accurate rules. Results show that the control heuristic may exist for subjects, yet that it can be brought under experimental control through the use of experimenter delivered instructions. Implications for a behavioral treatment of gambling and for a within-subject approach to the study of rule-governed behavior are discussed.
Article
The present study examined the degree to which slot machine near-miss trials, or trials that displayed 2 of 3 winning symbols on the payoff line, affected response times and win estimations of 12 recreational slot machine players. Participants played a commercial slot machine in a casino-like laboratory for course extra-credit points. Videotaped sessions were later coded to assess trial types and participant response latencies and win estimations for each participant. Results show that all 12 participants emitted verbalizations suggesting near-miss trials were of a higher approximation to a win than non-near-miss losing trials. Of the 12 participants 8 also demonstrated higher response latencies following losing trials than following winning trials. Variations across participants' response latencies were attributed to the presence of and response to near miss trials. The implications of the near-miss on game preference, resistance to extinction, and the development of a behavioral treatment for pathological gamblers are discussed.
Article
The purpose of the present study was to examine the participation of a variety of variables in gambling behavior. Forty-five subjects were exposed to an experimental roulette game played for course extra-credit points. In Phase 1, subjects played a fair game (i.e., completely random outcomes) without any instructions regarding the programmed contingencies for 75 trials. During Phase 2, subjects were exposed to an adjusted game with payback percentages of either p = .2, p = .8, or continued at p = fair and instructions related to play. Lastly, during Phase 3 all subjects were exposed to an adjusted game with a payback percentage of .2 and an option for escape. Results show that subjects exposed to inaccurate rules regarding play took higher levels of risk, made larger bets, and chose the escape option later than subjects exposed to accurate or no rules related to play. Although reinforcement density varied across subjects exposed to the same rule condition, there was no significant effect of that manipulation on subsequent gambles.
Article
According to Griffiths (1999), monetary rewards may only play a limited role in the maintenance of gambling behaviour. Many other factors, including near-misses, excitement, and social rewards should be taken into account when trying to explain persistent gambling. In this paper, it is argued that this eclectic (biopsychosocial) approach needs to be applied with caution. This is because of the absence of clear operational definitions of these other reinforcers (e.g. ‘near-misses’, social rewards); the fact that these reinforcers are likely to be related to the receipt or anticipation of monetary rewards (Walker, 1992); and the danger of over-explanation. If reinforcers can be direct or indirect, monetary or non-monetary, are there any aspects of gambling (apart from clear losses) which are not reinforcing?