“Lenny didn’t let his mind screw him up. The physical gifts required to play pro ball were, in some ways, less extraordinary than the mental ones. Only a psychological freak could approach a 100-mph fastball aimed not far from his head with total confidence.”- Michael Lewis, Moneyball
The simulation only lasts five minutes. That’s what Jordan Muraskin kept saying. Five minutes and you can be back out with your teammates, taking batting practice. Five minutes and you can slip that funny-looking thing off your head, the console that looks like a luminescent swim cap. Five minutes and your athletic career could be changed forever.
The baseball player from Bradley University still stared at him with a crossways look that cried, Just what is this supposed to be about?
A Dell Latitude E5440 laptop sat on the table. The screen was blank white. In a few seconds, a countdown will appear, Muraskin said. It will indicate which type of pitch to expect: fastball, slider or curveball. The pitch is just a green dot that darts straight or swerves, depending on the pitch. If the pitch is what you were told to expect, you press the button, J, to swing. Simple.
There were no seams, no release point, no pitcher, stands or dappled sunshine streaming through any clouds above. The simulation, administered late last October, indeed lasted only five minutes, and the player returned to batting practice as promised. But in that time, the researchers — Muraskin and his lab partner, Jason Sherwin — got all the data, brain data, they were hoping to acquire. And soon, too, did Bradley’s baseball coaching staff.
Mental analytics may be the next frontier in sports, and any number of companies has already set out exploring new ways to conquer it. Sherwin, 32, and Muraskin, 30, are younger and joined the party later than most of them. Neither had any background in business or baseball. Sherwin, who has floppy auburn hair and a scruffy beard, is the son of a conservative Chicago rabbi. Muraskin, more boyish-looking, with dark features and prominent eyebrows, is a skilled computer programmer. When they met, in the biomedical engineering department at Columbia University, Sherwin was studying the neural composition of cellists. Muraskin stumbled into neuroscience by researching Alzheimer’s and aging, and had been analyzing the efficiency of magnetic resonance imaging (MRI).
But as desk mates in the lab, they discovered that their independent studies into the process of decision-making dovetailed with a far different brand of research. What role the brain plays in hitting a baseball — the act that Ted Williams deemed “the hardest single thing to do in sport” — has been riddling scientists since before Babe Ruth was born. Late advancements in neuroscience research have permitted deeper inspections into the brain and its intricate network for interpreting 95 mph pitches than ever before. Sherwin and Muraskin elected to take a crack at it. Now they believe they have built a bridge between exploratory findings in the laboratory and real results on the baseball diamond.
When the two researchers hopped on a conference call with Bradley’s coaches in early November to discuss their findings, they finally got a sense of the scope of the concept they were developing, and what it could mean for baseball. The coaches immediately requested the results of one player who seemed to struggle at the plate, despite obvious athleticism and a picturesque swing. Muraskin patiently explained the metrics he had recorded, seven data points in all, categorized by names like Neural Decoding Performance, Decision Position Metrics and Neural Discrimination Strength, all of which were different ways of measuring how and why a player could hit a baseball or not. On his laptop, the readout was filled with line graphs, column graphs, data tables and heat maps.
The player’s neuronal curve was shifted backwards, Muraskin explained. In other words, he was late recognizing certain pitches and therefore late in deciding whether to swing.
The silence on the other end of the line told the scientists all they needed to know. Finally, one of the coaches said, “We never understood why he’s not the best player on the team.” Now they had a clue.
“It was like ‘Yes, yes, yes!’” Muraskin said later.
“Those things keep you paddling,” he added, “instead of stabbing the raft.”
As the manager of the New York Yankees, Joe Girardi lugs a large binder full of various charts and averages into the dugout with him for every game. The numbers might show that a particular left-handed batter hit fastballs to left field 26 percent of the time and might inform Girardi’s decision-making. And yet those numbers cannot offer any real insight into whether that outcome is intentional (the batter prefers going the other way) or the consequence of late decision-making or poor pitch recognition.
Regardless of what the so-called Moneyball effect has had on baseball, many of the game’s jazzy new offensive statistics available to managers like Girardi — like batting-average-on-balls-in-play (BABIP), and weighted on-base average (wOBA), swing percentages and other — still share one thing in common: they measure events that occur after the batter decides to swing (or not swing) his bat.
Muraskin and Sherwin’s model — using the simple video simulation they created and a basic electroencephalogram (EEG) they bought online — is aimed at a better understanding of why the statistics in Girardi’s binder read as they do. By measuring, assessing and, eventually, nurturing the brain activity of hitters before they swing, they believe those tiny fractions of seconds might differentiate a middling collegiate hacker from the next Mike Trout. Ultimately, rather than judging hitters’ physical actions and apparent abilities by subjectively watching from the stands or on video, scouts could assess prospects based on data, on the unconscious brain waves that drive those on-field movements and decisions. Armed with such cognitive information, ballclubs could decide whether a prospect was worth working with, or even drafting at all.
It might sound like science fiction, but from their earliest trials at Bradley, after just one simulated pitch, Muraskin and Sherwin could produce graphs that pinpointed precisely when the batter decided to swing versus when he decided to take, along the timeline of the pitch, down to the millisecond. After a few more pitches, they could create graphs that showed the spectrum of response times based on different pitches; graphs that assessed the batter’s concentration level (based on eye movement and the flutter of brain activity) before the pitch is even thrown; and graphs that correlate to the part of the brain that is firing when decisions are made.
This data could prompt a new way of thinking about top hitters. Their research already suggests that experts can recognize certain pitches the same way automobile enthusiasts can recognize the make and model of a car even as it disappears out of sight. It is not unlike the way bird-watchers can detect a specific bird by an instantaneous flash of color or flight pattern, or how a chess master can visualize and interpret movements on a board.
This expertise derives from the strength of signals that spring out of select regions of the brain — areas known primarily for early visual processing and motor control, located well back behind the ears. Suddenly, there are clues for what to look for, what may be detected neurologically in the players with star-power versus those without it. Sherwin and Muraskin’s expectation is that scanning prospects’ brains with an EEG will someday become as rote and orthodox as a yearly physical.
The impact on the game could be profound.
Offered such information as to what happens in the mind of a hitter before contact, swings or strides could conceivably be tailored to move quicker or slower, or pre-pitch routines altered to enhance concentration. Brain exercises might be developed to target precisely what an individual batter needs to change. Or, once a player understands what his mind is doing as a pitch approaches, feedback may allow the brain to self-correct the problem on its own, akin to the way meditation helps relieve some chronic health conditions. In theory, that sweet-swinging Bradley player could condition himself to recognize pitches faster and more accurately, enabling him to reach his potential. He might become the next Mike Trout instead of the kid who could not cut it in baseball after high school.
Similarly, scouts could suddenly target young prospects with phenomenal pitch recognition statistics, even if their on-field results or swing aesthetics had generated skepticism from scouts in the past, just as they now might do with a player with obvious but unfulfilled athleticism.
“You know the Disney movie, Million Dollar Arm,” Sherwin likes to say, referring to the film about two pitching prospects discovered during a talent contest in India. “We’re helping baseball teams find the million-dollar brain.”
In other words, find the brain, build the hitter.
Muraskin and Sherwin’s work arrives at an intriguing moment in baseball. After a bruising era of offense defined by sluggers and steroids beginning in the early 1990s and lasting until about a decade ago, pitching is now in vogue. Today, pitchers throw harder, with more violent movement than ever before, and as a result, teams are scrambling for ways to improve their offenses legally. The game is open to experimentation, and neuroscience just might be the next tool.
Muraskin and Sherwin are not the only scientists working in this direction. NeuroScouting LLC, an eight-year old startup in Cambridge, Massachusetts, with two neuroscientists at the helm, has devised computer software that can purportedly train and improve hitters’ visual and motor responses toward incoming pitches. Another company, Axon, has already partnered with Easton bats to create an iPad app said to “accelerate the mental skills needed to succeed as a batter.”
Both examples feature simulations that are said to work like popular mobile apps such as Lumosity and CogniFit, and neither company appears to use electrode caps specifically to profile the brain. But a few baseball teams, such as the Boston Red Sox, Chicago Cubs and Tampa Bay Rays, slowly have begun to adopt these brain-gaming techniques aimed at improving their batters’ skills. Some already incorporate videogames into the daily routines of prospects in their farm systems. According to the Wall Street Journal, Red Sox outfielder Mookie Betts, a fifth-round draft pick in 2011, quickly rose through the minor league system, in part, because of how well he performed in daily gaming drills. At age 22, he is now considered one of the best young players in the game.
Evidence is more than just anecdotal. Researchers at the University of California-Riverside published a study in the February 2013 issue of Current Biology that claimed the vision of baseball players on its Division I team improved after approximately 12.5 hours playing a specially designed video game. Those results then translated onto the field — the trained players struck out 4.4 percent fewer times than their untrained counterparts. UC-Riverside coach Doug Smith heralded the research, saying his players approached the plate with more confidence. Players believed they could also read, drive and watch television with more acuity as a result of the gaming.
Muraskin was intrigued, but had some questions about that report, and the overall efficacy of brain games as shortcuts to enhanced memory or athletic improvement. He and Sherwin chose instead to build their company — named deCervo, a combination of “decision” and “cerebrum,” the Latin word for brain — around a data-mining approach: By keeping their simulation as basic and authentic to baseball as possible, they believe the data figures they generate will better correlate to genuine baseball issues. And by delivering their neuronal research in understandable terms, such information will prove more useful, and teams more likely to accept the results and act upon them.
This spring, Muraskin and Sherwin visited four spring training complexes, carrying simply a laptop and a black aluminum briefcase, which contained the EEG. For the first time, they were testing professional players. A problem quickly arose, underscoring the need to test their methods in the real world and adapt accordingly — how do we translate these results into Spanish?
“I found a quote from Paul DePodesta,” Muraskin said, referring to the Mets’ vice president and one of the principle characters in the book, Moneyball. “He was giving a talk at Citi Field maybe two years ago, and somebody was tweeting about it. I think it was a field day for a tech company. And he said, ‘The problem isn’t with scouts or scouting. The problem is that it is based on a metric that is subjective, and not data-based.’
“What we’re trying to do [with our neurological data] is go right into there and say, ‘We’re scouting purely on the stats.’”
In 1993, while drinking beer and smoking cigarettes at a restaurant during spring training, Phillies outfielder John Kruk suddenly found himself being reprimanded by a stranger who thought his habits were unbecoming of a professional athlete. “I ain’t an athlete, lady,” Kruk said. “I’m a baseball player.”
Kruk hit .316 in 1993, and the line later became the title of his autobiography. Although he may not have been the trimmest ballplayer, and his range at first base was limited, the Krukster could hit. In 10 major league seasons, he was a three-time National League All-Star, with a lifetime average of exactly .300.
Understanding the biomechanics required to strike a 90 mph pitch has been a decades-long pursuit. As early as 1921, psychologists tested the sensory-motor skills of Babe Ruth, hoping to discover what made Ruth such a phenomenal hitter. The researchers asked him to tap a metal plate as many times as he could in one minute, or recognize flashes of letters displayed before him. Not surprisingly, Ruth’s aptitude in the tests was said to be exceptional. It is difficult to gauge how he measured up against other ballplayers because no one else was tested. Still, the study was an early, albeit tentative and flawed step in trying to determine why some players can hit, and others cannot.
Hitting continued to pique the curiosity of scientists. Two sports psychologists, Alfred W. Hubbard and Charles N. Seng, explored the role of vision during interceptive actions in 1954 and found that batters actually could not physically track the ball all the way to the point of contact with the bat. Three decades later, however, A. Terry Bahill and Tom LaRitz discovered that major league players could track the ball farther along than college players. A year later, Patricia DeLucia and Edward Cochran demonstrated the importance of peripheral vision in tracking the flight of the ball. Steven Radlo, Christopher Janelle, Douglas Barba and Shane Frehlich found that elite players could identify a fastball versus a curveball better than novices in 2001.
All were convincing studies, but they hardly explained what distinguished Mickey Mantle from Mickey Tettleton. Arguably, the most instructive piece of baseball literature for hitters came from one of the best to ever play, Ted Williams. In his seminal book, The Science of Hitting, published in 1971, Williams described his approach at the plate, ultimately revealing that it was not quite as scientific as the title suggested. Williams had exceptional vision and a sharp memory; he wrote that he could recall everything about his first 300 home runs — the pitcher, the count, the pitch itself, and where the ball landed. But at the plate, he was a “guess” hitter, surmising what pitch would be thrown where depending on the count and the situation on the field, which he could deduce because of his knowledge of pitchers’ tendencies.
Hitters found this understanding into Williams’ mindset invaluable, but there was little practical insight as to how to reproduce the results that enabled Williams to bat .406 in 1941. Williams could not explain how Kruk (and his diet of beer and cigarettes) could bat .320 at the major league level while Michael Jordan, arguably the world’s greatest athlete, could barely crack .200 with the Double-A Birmingham Barons in 1994.
So Dr. Harold Klawans, a Chicago-based neurologist, attempted to explain it neurologically. He found that the brain’s ability to learn to hit a baseball likely needed to take place during a critical period early in one’s life when the brain is still developing. But later, after the brain development has ended and neuronal networks are in place, Klawans argues the skill will be harder if not impossible to master, the way learning a second language becomes more rigorous as we age. The synapses between nerve cells are reinforced if used early, while the brain is developing, and both Williams and Kruk spent an inordinate amount of time hitting before they became adults. But left unused, these synapses atrophy and disappear.
“The sad fact,” wrote Klawans, “was that, at age 31, Michael Jordan’s brain was just too old to acquire that skill.” At that critical, earlier time, he had been shooting jump shots, not swinging at baseballs. By the time he picked up a bat as an adult, it was too late.
This “window” of skill acquisition underscored just how difficult the act of hitting a baseball is for the brain to master and enhance. There is no such thing as a baseball-hitting muscle that some players have and others do not. Players can point to their eyesight as a mark of distinction, but that alone does not guarantee success at the plate. When Louis J. Rosenbaum examined the vision of players on the Los Angeles Dodgers in 1993, half the team had visual acuity of 20/10. That was impressive, considering the theoretical limit of human vision is 20/9. But it was not as though half the Dodgers were on the All-Star team. And neither is everyone with 20/10 vision a major leaguer.
Pure reaction time among baseball players has never been clinically proven substantially better than average either, despite what researchers said about Babe Ruth. There is, however, a strong predictive element required in hitting, illustrated nicely by David Epstein in his book, The Sports Gene. He wrote about the 2004 Pepsi All-Star Softball Game, an exhibition featuring MLB All-Stars such as Pujols and Mike Piazza attempting to hit Jennie Finch, the Team USA softball pitcher.
Attempting was the operative word. Finch struck out Pujols as part of the pregame festivities. During the live game, she entered to face Piazza and Brian Giles, an All-Star outfielder for the San Diego Padres. She downed them both without as much as a foul tip.
The reason had little to do with velocity. Finch’s pitches, delivered from a mound 43 feet away, equated to a fastball thrown at 95 mph — hard, but nothing extraordinary to a major league hitter. Rather, it was the underhand angle from which Finch released the pitch that Pujols and Piazza found discomforting. It was like nothing they had faced before. Their database of knowledge from years of practiced learning was useless. They had no special ability to determine where her underhanded pitches would travel.
In his PhD studies while at Georgia Tech, Sherwin examined another complex form of decision-making, the kind faced daily by airplane pilots, who must monitor a dashboard of gauges, monitors and instruments in order to keep their planes steady. This required situational awareness, or what is called System 2 thinking, based off the work of Daniel Kahneman, the Nobel Prize-winning economist. These are slower, more deliberate forms of logic. A few years later, at Columbia, Sherwin began post-doctoral research on System 1 thinking, or rapid decision-making — hear a sound, make a decision.
It was a natural pivot for Sherwin, a skilled pianist and classical composer whose feet and fingers seemed always to be moving to some sort of silent beat. One of Sherwin’s earliest experiments involved professional cellists. A subject sitting in a dark room would listen to a clip of music, and then the harmony would be slightly modulated, say, from a G-major to a G-flat. Even non-musicians can generally notice that and click a button when it occurs. But in real musicians, the neurological reaction was different. Instinctively, their brains activated motor responses before the subjects consciously noticed the change in tone.
“In particular, it was more lopsided to the brain that controls their left side and arm,” Sherwin said. “That’s the hand position on the chord.”
The cellists wouldn’t raise their arm physically, but mentally, they wanted to. Theirs was a far stronger impulsive reaction than that of non-musicians, and the EEG could detect where and when the signals were coming from within the brain.
The insight into experts versus non-experts in music was a revelation. When Sherwin presented it to his lab for feedback, in the fall of 2011, Muraskin approached him afterward.
“He said to me, ‘Hey,’” Sherwin said, “‘Do you think we might see this with athletes?’”
Sherwin and Muraskin first combined their skills and interests to analyze ballplayer brains using an EEG in the spring of 2012. At the time, only a handful of other published experiments had related actual neural data to baseball. But, as they examined six random subjects, none with any advanced baseball experience, the Columbia researchers noticed something peculiar. Every time the batter incorrectly identified a pitch, there was a large neuronal current source — like water bubbling up from a spring — in a region of the brain called Brodmann area 10 (BA10), located in the prefrontal cortex, directly behind the forehead.
They wanted to probe further. Muraskin’s expertise using fMRI, which investigates neural activity by measuring changes in blood flow through the brain, allowed them to use a second tool to delve deeper than anyone else had tried. Running both pieces of equipment simultaneously was technically tricky —metallic objects do not interact well with MRIs — but they managed to do so. Participants still played the same simulation, and as pitches zipped in researchers could track stimulation in more distinct brain regions, such as the lingual gyrus, the lateral occipital cortex, and other parts of the occipital lobe and remaining cerebral cortex. And, again, when players did not recognize the incoming pitch, they detected activity in BA10.
They noted this phenomenon in a research paper, “A System for Measuring Neural Correlates of Baseball Pitch Recognition,” which they submitted in a competition at the 2013 Sloan Analytics Conference, a popular annual forum dedicated to sports analytics. The paper did not win, but Sherwin and Muraskin were exhilarated by the critical reception. They were sure they were on the right track toward determining which brain areas are important for accurate pitch recognition.
“It was totally encouraging,” Sherwin said. “I don’t think we really had any interest in winning or losing or anything. You just want to get there.”
Paul Sajda, the director of the Laboratory for Intelligent Imaging and Neural Computing at Columbia, where Sherwin and Muraskin met, told me that he and two researchers have since peeled back even more evidence of the neural science at play. Recently, using an eye-monitoring mechanism and pupilometer, they have begun tracking decision-making closer to its source: the brainstem, which produces neuromodulators like adrenaline, dopamine, serotonin and norepinephrine, a chemical related to arousal.
When a big fat pitch comes right down the middle, the brainstem activates, or arouses. Norepinephrine, generated in the locus coeruleus, a small pocket in the brainstem located just behind the ears, comes shooting into other areas of the cortex. That is what the EEG is detecting — brain arousal. Unlike a poker player, who must suppress a response when dealt a good hand, baseball players don’t fight the response. They act on it, and their motor functions are thought to have formed neural connections with the fusiform gyrus, the part of the brain involved in such traits as automobile, bird and facial recognition. Expert hitters produce similar neural responses when they recognize certain pitches, and their responses are generally far more dramatic than novices.
Why is this important? Localizing this could potentially be one way to separate the future stars from the future busts, Reggie Jackson from Mighty Casey. Said Sajda: “This is where the excitement comes from.”
In their business, Muraskin and Sherwin use only the portable EEG to test players, rather than dragging major leaguers into an MRI scanner. Though the neural picture they can receive is not as complete as with the fMRI, the data is still useful.
Naturally, I wanted to take a crack at the simulation and see what my brain might score. Maybe I was a latent star with a simple, correctable deficiency (I doubted it). When I visited in early February, their office was simply a desk in Columbia’s “Startup Lab,” a shared working space in Soho (they have since moved to a more private office about eight blocks east).
I was quickly ushered into a glass-walled conference room Muraskin had reserved for the afternoon. Opening a black briefcase, he pulled out a device about the size of a radio scanner. This was the amplifier, held in place on the base of my neck by a black elastic bandana wrapped around my head.
The electrodes were not the suction-cupped things I was expecting. They came already connected to each other on a clear pliable sheet of plastic that looked a bit like a Christmas tree.
“Players don’t want to wear something that looks like a Hydra on their heads, with all these wires going everywhere,” Muraskin said. “They want small and sleek.”
Using a conductive cream to ensure that the electrodes formed a close contact with the head, Muraskin fitted the cap to my scalp. He dabbed my mastoid bones (behind my ears) with an alcohol swab and placed two of the electrodes dangling from the amplifier on each one. Then on went the other nine electrodes, all at once, right over the top of my head. The cap fit snugly, but nothing more constricting than a cold-weather running beanie.
That was it; I was connected. Muraskin flipped the on/off switch on the amplifier, which connected to the external syncing unit via Bluetooth, and loaded up a computer program called X-Series Basic. I felt no sensation whatsoever. All of this — the equipment, the program, the lightweight carrying case — was made by the company Advanced Brain Monitoring. They purchased it online for $10,000.
He flipped on the game itself, and it was surprisingly rudimentary. The idea was simply to decide whether to swing or not to swing. The ball is represented by a small green dot against a white background. The dot moves, depending on the break of the pitch, and enlarges as it approaches, according to its velocity. The mean pitch speed was only 78 mph, but some curveballs were slower and some fastballs were faster. Sliders moved side-to-side and the curveballs dived down at a 12-to-6 rate. Fastballs remained straight and actually appeared to be tailing upwards.
As the pitch is being loaded (i.e. the windup), the simulation will tell you which pitch to expect: fastball, curveball or slider. You then “swing” by pressing a designated letter only if that is the pitch that was ultimately delivered. If not, you lay off. Location is not a factor — every pitch arrives at approximately the same part of the plate (the perspective is from the catcher). The only variable from pitch to pitch is the type, the velocity, and whether it is thrown from a lefthander or righthander.
Seventy-eight mph is batting practice speed for college players, but for a novice, even someone who played baseball through high school, the velocity was startling at first. In my first trial of 90 pitches, I accurately decided whether to swing or not swing only 52.26 percent of the time — basically guessing. The curveballs looked like fastballs, until they dropped. The fastballs sped by. And the sliders? Forget it. I had trouble differentiating them from either pitch.
I did improve over my second trial, and finished with an accuracy rate of 75.56 percent on my final two tries. But my response time averaged 421 milliseconds —much longer than my experienced counterparts. I thought I was deciding as quickly as possible. Turns out, I was dreadfully slow. No Mike Trout here.
Muraskin and Sherwin are in the process of developing a more detailed and comprehensive simulation, and Muraskin showed me a prototype their intern, Elizabeth McNally, a Dartmouth student, had been developing. The screen flashed to a fairly realistic representation of Nationals Park, in Washington D.C., as viewed from the batter’s box. Flags waved, there were fans in the stands, and the batter’s eye in centerfield was a lush green. The ball projected at the user would be white with red seams. The hope is that these peripheral comforts will make the simulation more enticing and relatable for a player, rather than green dots and white screens.
The simulated setting, though, does not need to get more elaborate for the test to remain effective. Even from the basic 15-minute exercise I underwent, Muraskin could still show me my independent component analysis (a series of squiggly lines reminiscent of a heart monitor); my sliding window logistical regression (a line graph that broke down the precise milliseconds when my brain firmly decided to swing or not swing); a graphical breakdown of prepitch readiness (based on brain signals just before the pitch is thrown); and numerous other analyses, including how accurately I selected and reacted to each different pitch.
Neither was I Bryce Harper, who currently leads the major leagues in on-base percentage (OBP) and is among the leaders in walks. Several graphs looked like heat maps and offered a dugout’s view of how far along the pitch was on its path to the plate before I decided to swing or take. My decisions to take, it seems, came in the last 5 feet before the ball crossed the plate. That’s not going to earn me many bases-on-balls, unless I get lucky.
The most precious statistic was Swing vs. Take, the decision conflict that played out over the course of just a half-second. A graph could show me how accurate I was when I made my decision between, say, 500 milliseconds and 600 milliseconds (80 percent), and even broke it down by certain pitches. Experienced baseball players often record 80-90 percent accuracy within 400 milliseconds. A coach with access to data for a player outside that range might one day be able to address the shortcoming in a hitter’s ability to detect certain pitches. Or, in a bleaker example, a scout might immediately decide to disqualify you as a prospect, no matter what your batting average in high school or college might say. He can already tell you that you won’t hit professional pitching.
We scrolled through printouts from Brown University baseball players — another team Muraskin and Sherwin have worked with — and could instantly see the talent gradient even among teammates. Their decision-making times ranged as widely as the results of the 40-yard dash in the NFL combine. One player, a first baseman, had truly impressive reaction times (100 milliseconds faster than his teammates) as well as exceptional accuracy reading all three pitches. But his pre-pitch preparedness reading showed his mind wandered, something a coach might theoretically be able to address. Or armed with the information that his neural decisions are unusually quick, another player could go into his next at-bat knowing he can wait longer on the pitch or perhaps swing a different way - either pull the ball or go to the opposite field, depending on the situation.
“Maybe he gives himself more time to see the pitch,” Muraskin said. “Now he has information to make adjustments, see the pitch a little bit longer, and then swing.
“You can’t get that information from anywhere else.”
Major league teams are understandably skittish about revealing too much about new ideas that might help their ballclub. The recent revelation that the St. Louis Cardinals were under investigation by the F.B.I. for hacking into computers owned by the Houston Astros only underscores how deeply guarded and mistrusting the atmosphere can be in baseball today.
Of the four major league clubs that deCervo visited during spring training, only one agreed to speak about its relationship with Sherwin and Muraskin. Even then, the assistant general manager who called one evening in June asked to remain anonymous, because he was not authorized to speak about the club’s business dealings. Then, after a few minutes, he took his plea for obscurity another step further, requesting that any particularly telling quotes not even appear in print, perhaps an indication that this team liked what they had found and sought desperately to keep it private.
Although this official also quickly downplayed deCervo’s influence, calling it “fairly experimental,” he added that he found that deCervo’s method appeared to be “directionally correct.” It meant that he believed they are on to something.
Vince Gennaro, a well-known consultant to MLB teams, said during a recent interview that most organizations would almost certainly be dabbling more and more with neuroscientific analyses, if they have not done so already. If he were advising a team, he said, he would tell them to give these guys a call.
However, Gennaro raised a common concern. On the baseball diamond, in a game, millions of external factors are at play. How can they be accounted for in a controlled laboratory setting? In other words, as well as deCervo simulates pitches, it’s still not Clayton Kershaw standing 60 feet, 6 inches away on the mound, and there is no fear of getting hit by a pitch or disappointing a teammate by failing to move the runner over.
“It’s not an accusation,” said Gennaro, who is also president of the Society for American Baseball Research, which lent its acronym, SABR, to the forefathers of sabermetrics. “I just wonder how that changes everything. We’re dealing with the human mind. I just can’t help but feel a little awkward about separating that reality out from the laboratory.”
The club official that deCervo has been working with expressed similar concerns. He said he had not yet been convinced that strong results in a controlled experiment were actually the brain at work, rather than just the placebo effect (this team put about a dozen players through the test, all of them minor leaguers). Was the game actually stimulating a player’s mind or just building confidence? He needed the answer to that question.
As an evaluation tool, rather than one that promised to facilitate improvement at the plate, the official believed deCervo seemed to offer some promise. But again, the official expressed hesitancy. He wondered whether the data provided by the simulation was actually new and usable, or simply just a confirmation of suspicions they already had? If so, what is the value in that?
“The majority of the teams we worked with were more interested in the nuances here,” Sherwin said over lunch in early June. “Hitting is one of the hardest things to do in sports. There’s no monotonic score. It’s got a bunch of peaks and troughs. Some people are great bunters, some people are great opposite field hitters, some hit better late in the count, whatever. In general, we found that the mid- and small-market teams are more interested in the potential for future growth and then accessing that future growth, using this insight that you have now into the decision-making of the player.”
I asked Sherwin to respond to Gennaro’s critique about how real-world scenarios cannot be simulated by a test.
“He’s right,” Sherwin said.
He held out two hands over his plate separated by about a foot. “If this is all of the experience of hitting a baseball, standing in the box, we’re doing a part of that. Not all of it. The question is how much of it?”
“All those things at Yankee Stadium, bottom of the ninth, all that stuff,” Sherwin added, “those are all inbound stimuli on the nervous system. The question is whether the batter is able to take those inbound stimuli and just put them over here. Just focus on the pitch. Just like it’s any other pitch.”
Sherwin’s own mind went wandering then, as he hypothesized ways to pinpoint which stimuli can be the most distracting, then discover ways to eliminate them as factors, a step on the path, as he put it, to “orient somebody in the clutch.”
This is not something deCervo is yet capable of, but the notion is another branch on the giant sycamore of ideas that grows ever larger and denser in Sherwin’s head. Sometimes they lead to sudden insights. After the first day of working with teams in Arizona in March, Sherwin convinced Muraskin that their simulation needed a dramatic change. The pitch recognition simulation — fastball, slider or curve — was incomplete. Hitters just weren’t trying to recognize the kind of pitch to hit, but see the pitch and hit the pitch, only if it is a strike.
So the researchers wrapped up their trials at 10 p.m., returned to their hotel room and revised their game overnight, superimposing a rectangle representing the strike zone across the middle of the screen, one that faded out as the pitch approached. The batters would then be asked whether the pitch was a ball or a strike.
Their accuracy compared against an infallible, fictitious batter produced a metric they call the Perfect Eye Index. It is unlike any data they had generated previously, because it has nothing to do with neuroscience. It is simply an accuracy reading. The EEG might as well have been left unplugged.
“If it’s a slider, who cares?” Sherwin said. “If it’s a strike, that’s what matters [to hitters].” Sherwin concluded that since the batter’s objective is simply to reach base, whatever pitches he saw to get him there are tangential.
Adjusting on the fly was a good indication of deCervo’s conscientiousness toward its clients, even if it gave somewhat of a disorganized impression. They are still experimenting with proper balances, trying to marry what players want (a more realistic game) with the purpose of their company (gathering usable brain data). For the remainder of spring training, they split their exercises into two sessions: 20 minutes of pitch recognition, 20 minutes of balls vs. strikes. In reality, of course, the hitter in the batter’s box does both simultaneously.
One objection I heard from the baseball team official was that he thought neuroscience companies like deCervo would struggle unless they could effectively focus on one or two aspects of the game that could be distilled into some kind of app. Otherwise, they risked drowning teams - and players - with information and ideas, paralysis by analysis.
From his 10th floor office looking south across the rooftops of Columbia’s main campus, Sajda, who is an advisor to deCervo, said that though the founders have made fantastic strides, they are still relatively small steps, and much of their work and research cannot be validated for months and years down the road.
“The bottom line is we still don’t really know how the brain works,” Sajda said.
People have a right to be skeptical, he said. Companies are clambering to get a slice of the neuroscience market share without a complete understanding of what they are peddling and whether it really works for what they are aiming toward.
“It always comes down to defining the value,” Sajda said, adding, “If it works out, if it’s as good as people think it is, there’s a huge competitive advantage. So is it worth the risk?”
On a hot and sunny Arizona afternoon in early April, Sherwin and Muraskin settled into seats on a small set of bleachers between home plate and the home dugout, a view that offered them a good vantage point of both the pitcher and the batter in an exhibition between minor league clubs in extended spring training.
Already, they had spent the morning examining the players’ neural responses with simulation tests. Here they wanted to see the game in real time. After a while, the organization’s minor league hitting coordinator hopped into a seat next to them and they watched a batter. As the pitch flew in, the batter lunged out slightly over the plate.
“You see that?” the hitting coordinator said, as relayed by Sherwin. “That means he’s not seeing the ball well.”
Sherwin and Muraskin had not even noticed it. But the coordinator said it was something he had spent the last decade training his eyes to detect, among a plethora of slight mechanical quirks and flaws. The conversation turned from what the batter was doing to what he was not doing — seeing the ball all the way in, keeping his head still and his eyes focused as the pitch approached. Instead, because he was having trouble recognizing the pitch, his body induced a motor response (lunging) to better evaluate the trajectory or direction of the ball. This took up precious time and took him out of position.
Sherwin offered a quick evaluation: The batter’s perceptual side — his visual prediction capability — wasn’t accurate enough for him to know where the ball was going to go or whether he was going to hit it. So he adjusted his approach, something hitters do all the time. If they are slumping, or having difficulty seeing the ball, they might open their stance to direct both eyes at the pitcher until the pitch is coming, and then stride toward the plate. Coaches might advise them to try to increase their focus, settling on a specific target during the pitcher’s windup, as George Brett (his hat) and Steve Garvey (his face) said they would do, or they might choke up for better bat control, a tactic routinely employed by Tony Gwynn.
For most hitters, it’s trial and error, and experienced pitchers can pick up on many of these adjustments, and then attack accordingly. Brent Walker, a former college pitcher, said he could always tell when batters were struggling to read his pitches. So starting eight years ago, Walker, now president-elect of the Association for Applied Sport Psychology with a doctorate in sports psychology, devoted himself to finding ways for hitters to improve their ability to recognize pitches without tipping their hand. At first, he used a simple video simulation of a pitcher seen from the hitter’s perspective. Moments after the pitch was released, the screen would darken, and the batter would be prompted to record what pitch he expected to arrive.
“The idea was training,” Walker said.
Still, nobody knew what was happening internally, in the brain. Walker knew that some researchers were pinpointing alpha waves — otherwise known as the Berger rhythm — in the left hemisphere of the brain as clues to enhancing concentration in activities like golf putting, archery and pistol shooting. Still, Walker wondered about its application in the batter’s box. He remained skeptical that alpha waves could help you hit Kershaw.
But when he read Sherwin and Muraskin’s research for the Sloan Analytics Conference, Walker recognized a smart melding of the two brands of science. It did not hurt that they were Columbia researchers, and that in July of 2012 Walker had joined Columbia’s athletic administration staff as an associate athletics director. He emailed Muraskin and scheduled a meeting early last summer.
“There are some interesting possibilities out there based on what they’re doing,” Walker said. “They’re going to the next level of it.”
Walker, like the minor league hitting coordinator, had always had theories about hitters’ tendencies when they were not seeing pitches well. But suddenly, with some supporting neural data, he had a more concrete understanding of what was at play and how to change it. Maybe the coaches might throw a struggling batter more off-speed pitches in batting practice, or lengthen his swing. A hitter might improve through video simulation, or his brain might self-correct on its own, armed with new feedback. Instead of guessing at a solution to a problem — lunging at the ball or choking up — a hitter might know how best to fix it.
Sherwin said he saw the minor-league hitting coordinator’s eyes light up at the thought of having such a window into the hitter’s mind.
“It was kind of like how I explain to my grandmother how Tinder works,” Sherwin said. The art of dating — not unlike the art of hitting — had been boiled down into an algorithm where it was then reconstituted into something that could be done with the swipe of a finger.
“That was not recognized as a possibility before,” Sherwin said.
In baseball, there remains a sharp division between the adopters of new analytical approaches to the game — now more mainstream than ever — and those who want to keep scouting and training the way it was for Ted Williams and Mickey Mantle. Hitting coaches tend to fall on the older-school side of the spectrum, and those I spoke with were hesitant to give away all their jobs to video game companies.
“I think some vision training/pitch recognition devices that I have seen are helpful,” said Troy Silva, a Seattle-based hitting coach, author and former minor leaguer. “But most are gimmicks.”
Charley Lau Jr., a renowned hitting guru whose father, Charley Sr., honed the swings of Hall of Famers Reggie Jackson, Dave Winfield, Carlton Fisk and George Brett, said the only mental component he incorporates into his teaching currently comes as an amateur therapist, trying to free the batter’s brain of external distractions.
“I think the game is complicated enough,” Lau said. “Most people can’t pass Hitting 101. All of a sudden you’re taking hitting to a level where you need to be some kind of a genius to be a hitting coach, and there’s not many of them out there.”
Toward the end of our conversation, though, Lau recalled that as the hitting coach of the Kansas City Royals in 1971, his father introduced a novel piece of technology into the game that was seen as revolutionary. It was bulky and about the size of a portable television: a video camera. A generation ago, its use was also seen as radical and its effectiveness doubted. Today, its value is unquestioned.
Baseball, perhaps more than any sport, is always straddling that delicate line between old school and new. The tug-of-war wages on, deCervo wades in, and the march of science and technology continues pressing ever forward.
Sherwin remembers back to March 2013, shortly after they had published their first baseball-related research paper. When Sherwin, Muraskin and Sajda were invited to M.I.T. to present their early findings using EEG and fMRI to the Sloan Conference. As he walked through the main hall of presentations, Sherwin encountered booths of major league scouts exhibiting their new approaches to advanced analytics.
“They were talking about regression models and that kind of stuff they were doing, in the Moneyball vein,” Sherwin said.
Sherwin asked them if they were curious about digging down even further, to the cognitive level, to extract what exactly it was that might be generating that WAR number or strikeout-to-walk ratio.
“Most guys looked at me like I was crazy. They were like, ‘Nah, we don’t need to drill down to that level,’” Sherwin said. “I’m like, OK. I’ll see you in 10 years.”