Home > Pattern Recognition > Shoot to get hot, shoot to stay hot – or not?

Shoot to get hot, shoot to stay hot – or not?

Picture this: you’re watching your favorite professional basketball team on television
when suddenly their best player gets fouled again—that’s already the sixth time tonight, and he
hasn’t missed a free throw yet! You watch eagerly as he steps to the free throw line for the first
of two shots. He bounces the ball, once, twice, stares down the rim, and shoots—swish—a
perfect shot once again. The referee hands him the ball for his second and final shot while you
think to yourself, “There is no way he will miss this shot, he’s hot and having a great night. He
hasn’t missed a single free throw all night, and he just made the first shot, so this one has to go
in.” But your confidence is shattered when the ball leaves his hands and soars just a bit too far,
bricking off of the backboard

Nothing but air.

and into the hands of the other team. If this situation sounds
familiar to you, then you’ve fallen victim to what psychologists call the “Hot Hand Fallacy,” or
the erroneous belief that someone’s performance in a sporting event or similar life circumstance
is expected to occur in significant “streaks”—in other words, good outcomes are more likely to
occur in conjunction with other good outcomes, and, likewise, bad with bad.

Yes, that’s right, the erroneous belief, as there exists significant research that tells us that
the state of a player “being hot” is nothing more than a figment of our imagination. If you have
trouble believing this, then you’re not alone. Gilovich et al.’s extensive 1985 study found that
91% of college-aged basketball fans believed that one is more likely to make a shot after just
having made a basket as opposed to missing a basket. Furthermore, the participants, on average,
estimated a player was nearly 20% more likely to make a shot after having made one compared
to after having missed one (Gilovich, Tversky, & Vallone, 1985).

So, if this belief is so ingrained in people’s minds, how can it be wrong? Gilovich and company turned to statistics to disprove the perceived phenomenon. First, they looked at the Philadelphia
76ers’ 1980-1981 season shooting statistics, and searched for what they called “runs” or streaks
of either makes or misses (a streak is a continuous sequence of either makes or misses ). They compared the actual amount
of streaks a player had to the expected amount they would have had if their shots were
independent of one another, and found that of the 76ers, all nine players, shot nearly equal to or
lower than the expected number of runs, suggesting that there is no such thing as a shooter

The 76ers squad in question

getting “hot” (Gilovich, Tversky, & Vallone, 1985). They calculated this expected number by looking at the shooters’ shots made and shots missed, then assumed a random dispersion of makes and misses from there.
This study delved even further into the hot hand effect, analyzing free-throws in both
professional games and a controlled shooting experiment, and consistently found that basketball
shots were independent of one another (Gilovich, Tversky, & Vallone, 1985). So, how do they
account for the widespread, fallacious belief that players can have a “hot hand”?

To explain this common, allegedly erroneous conviction, the researchers turned to errors in pattern recognition. Pattern recognition is a cognitive process in which, by using information from memory, one is able to
make meaning of their environmental stimuli. In this pattern recognition system, bottom-up and
top-down processing work in tandem to allow for fast recognition of meaningful stimuli.
Bottom-up processing relies entirely on stimuli alone. If humans utilized only bottom-up
processing, then the “hot hand” effect would be non-existent, as we would only be able to
consider each shot independently without drawing on prior knowledge or experience. However,
top-down processing allows humans to do just that—it integrates conceptual knowledge and
expectations into the pattern recognition system. This is why the word superiority effect exists,
in which we are more easily able to recognize a letter when it appears in a word than when it
appears alone. In this case, top-down processing is immensely helpful in improving reading
comprehension and speed, but top-down processing is not all its cut out to be, as humans often
have a tendency to recognize meaningful patterns in everything. For an example, see face-
pareidolia, the phenomenon in which one sees faces in inanimate objects.

The Wolverines barely miss an easy basket…I bet the Mules could take them

In the case of the Hot Hand Fallacy, researchers conducting the 1985 experiments cited
top-down processing as the reason for erroneous conceptions of a “hot hand” (Gilovich, Tversky,
& Vallone, 1985). First, they reasoned that a fan’s intuition in a player being on a good or bad
shooting streak could be influenced by other facets of their game, such as their defensive play or
rebound total (Gilovich, Tversky, & Vallone, 1985). In this scenario, the other facets of a
player’s game are the prior experiences and conceptual knowledge that the fan will apply to the
stimulus, the player’s shooting performance. From there, the fan makes a prediction regarding
the next shot that is based off of many more factors than simply the player’s shooting ability.
The researchers also considered the common event in which the basketball nearly goes
through the hoop, but pops out or is just slightly off target. This is objectively a miss, a negative
event, however, the study found that if a player is perceived as being “hot,” then this close-call
missed basket is interpreted as a positive event, a continuation of the hot streak (Gilovich,
Tversky, & Vallone, 1985). Conversely, if the player is perceived as being on a “cold” streak, the
near-miss is considered to be validation and continuation of the player’s stretch of poor shooting
(Gilovich, Tversky, & Vallone, 1985). In this scenario, if the spectator used only bottom-up
processing, then the ball barely missing the hoop would be considered to be a negative event
regardless of circumstance. However, top-down processing would cause one to erroneously also
consider the previous shots leading up to the missed shot, which could lead to the interpretation
that a nearly-missed shot was actually a continuation of a “hot” streak (Gilovich, Tversky, &
Vallone, 1985). This manipulation of an outcome to match one’s expectations is a common phenomenon known as the expectation bias. Gilovich et al. claimed that, in reality, it was a miss in a far more random series of successes and failures (Gilovich, Tversky, & Vallone, 1985). In other words, they determined
that each shot was independent—its probability of going in was unrelated to the outcome of
previous shots—and our interpretation that the outcomes of shots is related was the result of
erroneous pattern recognition.

So, if this truly is a fallacious interpretation of random events, why does it occur? Researchers have proposed that the Hot Hand Fallacy is the result of evolution (Wilke & Barrett, 2009). They theorize early humans adapted to recognize patterns of clumped, related information as opposed to processing their environment as a series of random events (Wilke & Barrett, 2009). For example, consider foraging and hunting. When encountering food sources, it would be important to recognize where the most fruit tends to grow, where prey tends to graze, etc. If humans considered each hunt or foraging session as an independent event, then their search for food would be based entirely off of bottom-up processes, such as footprints or other sensory clues. However, the evolution of top-down processing allowed humans to recall prior knowledge and experience – they could seek out popular breeding grounds, recognize that animals need water and search near rivers, and learn what plants grow where and what animals graze where. By processing information as a meaningful series of patterned clumps, humans became much more efficient. However, you certainly may not feel more efficient when you realize that half of the sports predictions you’ve ever made have been based off of misplaced pattern recognition.

Now, there is more at stake here than just the intuitive predictions of fans and spectators.
Gilovich et al. also interviewed the players and coach of the same Philadelphia 76ers team about
the idea of a player having “hot hands.” After a series of questions—including “Is it important
for players on a team to pass the ball to someone who has just made several shots in a row?” to
which every member of the team responded “yes”—it was clear that coaches and players often
made crucial game-time decisions based on the Hot Hand Fallacy, which is based off of an error
made in the pattern recognition process.

Perhaps you are still skeptical. Perhaps you don’t buy their method, don’t subscribe to
their statistics, or simply do not agree with Gilovich et al.’s conclusion due to your own personal
experiences.

Well, you’re certainly not alone.

Gilovich et al.’s infamous 1985 paper, “The hot hand in basketball: On the misperception
of random sequences,” came under significant fire in the decades that followed its publication,
sparking the epic saga surrounding the “hot hand” effect.
In 2002, Kevin Korb and Michael Stillwell gave a presentation at the International
Conference of Cognitive Science critiquing Gilovich et al.’s approach to analysis, claiming that
it lacked statistical power (Korb & Stillwell, 2002). Statistical power describes a statistical method’s ability to recognize significant effects when significant effects do exist.

The next year, J.J. Koehler and Caryn Conley recognized the cognitive psychology
community’s concerns with Gilovich et al.’s statistical analysis methods, and designed a new
methodology that not only took these apprehensions into consideration but also analyzed
performances at the NBA Long Distance Shootout contest and identified when television
announcers deemed a certain player “on fire” (Conley & Koehler, 2003). They determined that
the original 1985 study, despite its lack of statistical power, was correct to conclude that the Hot
Hand Fallacy was indeed a fallacy.

It only took thirty years for someone to analyze a sport other than basketball

However, more research followed from Green and Zwiebel in 2014, when they
investigated the Hot Hand Effect in baseball and found evidence for its existence in ten statistical
categories, such as RBIs or Home Runs (Green & Zwiebel, 2014). They believe that spectators are correct to use their top-
down processing and prior knowledge in the prediction of short-term performance of athletes,
but that the sport of basketball, due to the existence of defensive pressure, is an inappropriate
setting to test for a Hot Hand Effect (Green & Zwiebel, 2014). They argue that prior literature has falsely categorized the Hot Hand Effect as a mistake, when in reality it is a viable, necessary adjustment made to account for equilibrium (Green & Zwiebel, 2014). This
suggests that pattern recognition is essential for the human predictive abilities because it allows
one to adjust for equilibrium, to notice trends in performance and respond accordingly. Another
study also found evidence of a Hot Hand Effect, but this time in basketball, returning to the roots
of this problematic cognitive dilemma and offering the saga a satisfying conclusion (Miller &
Sanjurjo, 2016). It did so through more powerful statistical methods. Decades after the original study, it seems as if we’re actually correct to perceive athletic performance in “hot” or “cold” streaks.

The saga of the Hot Hand “Fallacy” has been a landmark in cognitive psychology. For decades, the findings of Gilovich et al. stood uncontested, temporarily proving human intuition to be at fault. However, modern researchers have proven why no psychological discovery can ever be set in stone. Through new statistical methods and a broader range of analysis, it seems that humans are actually correct to identify patterns in seemingly random sequences.

References:

Gilovich, T., Tversky, A., & Vallone, R. (1985). The hot hand in basketball: On the misperception of random sequences. Cognitive Psychology.

Green, B., & Zwiebel, J. (2017). The Hot-Hand Fallacy: Cognitive Mistakes or Equilibrium Adjustments? Evidence from Major League Baseball. Management Science, https://doi.org/10.1287/mnsc.2017.2804.

Koehler, J. J., & Conley, C. (2003). The “hot hand” myth in professional basketball. Journal of Sport & Exercise Psychology, 25, 253-259, https://doi.org/10.1123/jsep.25.2.253.

Korb, K. B., & Stillwell, M. (2002). The story of the hot hand: Powerful myth or powerless critique? Presented at International Conference on Cognitive Science.

Miller, J. B., & Sanjurjo, A. (2016). Surprised by the Gambler’s and hot Hand Fallacies? A Truth in the Law of Small Numbers. IGIER Working Paper No. 552, http://dx.doi.org/10.2139/ssrn.2627354.

Wilke, A., & Clark Barrett, H. (2009). The hot hand phenomenon as a cognitive adaptation to clumped resources. Evolution and Human Behavior, 30, 161-169, https://doi.org/10.1016/j.evolhumbehav.2008.11.004.

Images:

Image 1: https://www.usatoday.com/story/gameon/2013/04/04/steve-nash-dwight-howard-free-throws/2052547/

Image 2: http://www.tradingcarddb.com/ViewCard.cfm/sid/25504/cid/2987177/1980-81-Topps—Team-Posters-12-Philadelphia-76ers

Image 3: https://www.google.com/search?biw=1536&bih=735&tbm=isch&sa=1&ei=JGfiWtCQLM67ggeO5qfQDw&q=basketball+rolls+around+rim+gif&oq=basketball+rolls+around+rim+gif&gs_l=psy-ab.3…5632.9501.0.9625.14.12.2.0.0.0.133.898.7j3.10.0….0…1c.1.64.psy-ab..2.5.373…0j0i8i30k1j0i24k1.0.grzIYFcjX_c#imgrc=zT43lcS0sDaPKM:

Image 4: http://www.masnsports.com/steve-melewski/2018/03/walk-off-win-adam-jones-homers-in-the-11th-for-the-win.html

  1. swgray20
    May 15th, 2018 at 23:29 | #1

    This post was very interesting and the lines of research that you discussed allowed me to navigate through the history of the “hot hand fallacy”. The role of top-down processes is a clear driving factor of this cognitive bias, and I thought it was particularly interesting that you discussed evolutionary forces that might explain our reliance on top-down, contextual information in pattern recognition. A lot of the previous research on this phenomenon appears to be primarily statistical and looking at trends in how people perceive “hot” and “cold” streaks and whether these assumptions are valid. I wonder how these results would match up against a sample of fans, assessed for their general knowledge about the sport they were watching, asked to predict a player’s later performance in a game based on their earlier performance. An experiment could be set up in which participants are asked to rate second half performance based on their observations from the first half. I’m interested in looking at the potential role of attention as well as emotion (towards a particular team or player) and potential features of a player’s performance that would lead to a more robust “hot hand fallacy”. The way in which fans identify patterns and determine whether or not players are “hot” may be more complex than analyzing their statistical contributions. I think that this effect could also be mediated by a person’s emotions towards a player or team that is objectively “hot” and also whether a person was attending to a player at specific instances that would deem his current streak “hot”.

You must be logged in to post a comment.