Authors: Dean Buonomano
Imagine separating twins at birth and having one baby be raised by newlyweds who teach at the local high school, and the other by the Pirahã, the hunter-gatherer tribe of the Brazilian Amazon. Five years later one of the twins will know the adventures of
Dora the Explorer
by heart, how to operate a cell phone, and how to speak very good English; the other will know how to fish, swim, and will have mastered what may be the most difficult of all languages. Two decades later the first might be in graduate school, coming to terms with special relativity, while the other uses his considerable skills to provide food and shelter for his family. Despite the vastly different computations being performed, both twins would be using the same out-of-the-box computational device, without any need for an external agent, such as programmer, to develop and install the English or Pirahã software packages—culture is the programmer. This is why the brain can be said to be an open-ended computational device. Yes, it is constrained by the boundaries laid down by our neural operating system: we will never be able to manipulate numbers with the accuracy and speed of a calculator; we may always have memory flaws and capacity limits, an outdated fear module, and a vast collection of cognitive biases. But, still, the human brain stands alone in its ability to adapt to unforeseen environments and grapple with problems evolution never anticipated.
The brain is defined by its ability to change itself. The hopelessly complex tangle of axons, dendrites, and synapses within our skulls do not form a static sculpture, but a dynamic one. Our experiences, culture, and education rewire our neural circuits, which in turn shape our thoughts, actions, and decisions, which in turn alter our experiences and culture. Through this infinite loop we have advanced both the average amount of time each of us inhabits the planet and the quality of our stay. We overcame many of our prejudices, and at least in principle have come to accept that every individual is entitled to the same rights and freedoms. Despite the brain’s inability to store and manipulate large numbers we have devised machines to perform these computations for us. We have advanced beyond the stage of offering human sacrifices to gods we have created in our own image. Although smoking continues to be a serious health threat, fewer young people begin smoking as a result of educational campaigns. And even a little bit of skepticism and common sense go a long way toward protecting ourselves from blatantly misleading advertising and political demagoguery.
Over the millennia our conscious reflective system has bootstrapped itself to a singular stage: one that has allowed the brain to narcissistically peer into its own inner workings. As this inner journey progresses we will continue to unveil the causes of our many failings. But like those of us who resort to setting our watches ahead five minutes to compensate for our perpetual tardiness, we must use our knowledge of neuroscience and psychology to teach ourselves to recognize and compensate for our brain bugs, a process that no doubt would be accelerated by teaching children about the strengths and flaws of their most important organ. Given the pervasiveness of the brain’s flaws, and the increasingly complex and ecologically unrealistic world we find ourselves in, embracing our brain bugs will be a necessary step toward the continued improvement of our own lives and the lives of our neighbors near and far.
I suspect I owe my fascination with the inner workings of the brain to my baby sister. The brain’s voyage from puzzled babyhood to dexterous adolescence leaves an indelible mark on anyone who witnesses that transformation. I thank my sister for her unknowing participation in a few of my early harebrained “studies,” and for her later enthusiastic encouragement of my slightly more sophisticated forays into neuroscience.
One of the points of this book is that human memory is not well suited to store certain types of information, such as names. So in an effort to decrease the amount of information that the majority of readers may not need to know, I sometimes omitted the names of the authors of studies from the main text. In the endnotes, however, I made every effort to attribute the findings to those scientists who are primarily responsible for them, but I apologize in advance for those instances in which I failed to give credit where credit is due.
It is an unfortunate fact of science that not all scientific findings prove to be correct in the long run. Progress in science requires that multiple independent groups eventually replicate the findings of others. Initially exciting findings are sometimes ultimately proven to be incorrect, having been the result of statistical flukes, methodological oversights, poorly executed experiments, or even fraud. For this reason, to the extent possible, I attempted to limit the findings discussed to those that have already been replicated; and in an effort to convince myself and the reader of the veracity of a finding I attempt to cite more than one paper to substantiate the results in question. This is not to say that some of the topics and ideas presented are not highly speculative in nature—particularly attempts at linking psychological analyses of behavior to the underlying mechanisms at the level of synapses and neurons, as in the discussion of our susceptibility to marketing. But I have attempted throughout to convey what is accepted science and what is scientific speculation.
This book would not have been possible without the help of a multitude of friends and colleagues. Their roles in this book take many forms: educating me on some of the material covered, reading one or more of the chapters, or simply not mocking my questions. The following people fall into one or more of these categories: Jim Adams, Shlomo Benartzi, Robert Boyd, Harvey Brown, Judy Buonomano, Alan Burdick, Alan Castel, Tiago Carvalho, Michelle Craske, Bruce Dobkins, Michael Fanselow, Paul Frankland, Azriel Ghadooshahy, Anubhuthi Goel, Bill Grisham, April Ho, Sheena Josselyn, Uma Karmarkar, Frank Krasne, Steve Kushner, Joe LeDoux, Tyler Lee, Kelsey Martin, Denise Matsui, Andreas Nieder, Kelley O’Donnell, Marco Randi, Alexander Rose, Fernanda Valentino, Andy Wallenstein, Carl Williams, and Chris Williams. I’d especially like to thank Jason Goldsmith for his thorough comments on much of the manuscript and his many stimulating suggestions.
I would also like to express gratitude to my friends who over the years have generously shared their time, knowledge, and ideas, and nurtured my scientific meanderings. These include, but are not limited to, Jack Byrne, Tom Carew, Marie-Francoise Chesselet, Allison Doupe, Jack Feldman, Steve Lisberger, Mike Mauk, Mike Merzenich, and Jennifer Raymond. My own research has benefited from the support of the National Institute of Mental Health and the National Science Foundation, as well as from the support of the departments of neurobiology and psychology at UCLA.
I am grateful to Annaka Harris, my editors Laura Romain and Angela von der Lippe at Norton, and my agent Peter Tallack for their guidance and editorial expertise. Additionally, I am indebted to Annaka and Sam Harris for their invaluable advice and encouragement throughout every developmental stage of this book.
I thank my wife, Ana, who not only indulged my wish to write this book, but provided the support and environment that allowed me to complete it. Last, and most of all, I’d like to thank my parents for their nature and their nurture.
INTRODUCTION
1
Proctor, 2001.
2
Tversky and Kahneman, 1981; De Martino et al., 2006; Berger et al., 2008.
3
I would like to, rather self-servingly, use the term
brain bugs
to refer not only to the cognitive biases (Chapter 7) but also our memory flaws, susceptibility to advertising and fearmongering, and our propensity to subscribe to supernatural beliefs. In short, any and all aspects of human behavior that can lead to irrational and detrimental behaviors and decisions. Of course, as will be discussed in depth, the same aspect of cognition can be beneficial in some contexts and harmful in others (computer bugs can be harmless in most situations but problematic in others). Piattelli-Palmarini has used the term
mental tunnels
to refer to the cognitive biases we are subject to (Piattelli-Palmarini, 1994). Brown and Burton (1978) have used the term
bugs
to refer to the types of addition and subtraction errors made by children. Robert Sapolsky has also written an article on “Bugs in the Brain”; however, the term was used literally to refer to parasites that live in the brain and influence behavior (Sapolsky, 2003).
4
McWeeny et al., 1987; Burke et al., 1991.
5
This procedure for studying memory errors is referred to as DRM (Roediger and McDermott, 1995).
6
Michael Luo, “Romney’s slip of tongue blurs Osama and Obama,”
The New York Times
, October 24, 2007.
7
CAPTCHAs are not really a Turing test, but can be thought of as a reverse Turing test that allows computers to positively identify humans. The advantage of CAPTCHAs is that they provide a rapid, objective, and easy-to-administer test.
8
Basing CAPTCHA on the analysis of pictures raises the problem of cataloguing exactly what is in a picture to determine if the answer is right or wrong. This problem may be solved during the test by asking individuals to interpret an already catalogued picture and a novel one during each test, and having the novel one interpreted by many different individuals. By cross-referencing the answers across multiple individuals one can set the correct answers automatically. Using increasingly more complex tests, it is likely that we will continue at least for a while to be able to generate litmus tests that only humans can pass.
9
Turing, 1950.
10
Stanislas Dehaene’s book
The Number Sense
(1997) offers a superb discussion of the numerical skills of humans and animals, as well as a glance at the extreme ranges of mathematical abilities of humans.
11
Gifted mathematicians often report developing an affinity for specific numbers, and each number may have a certain personality. For example, that the number 97 is the largest two-digit prime, or that 8633 is the product of the two largest two-digit primes. However, it does not appear that they have a specific intuitive feel for the distinct quantitative difference between 8633 and 8634, in the same way we do for the numbers 1 and 2.
12
Four.
13
These number are admittedly merely estimates. The value of 90 billion neurons comes from a recent study based on cell fractionation (Herculano-Houzel, 2009). The estimate of 100 trillion synapses comes from studies suggesting that on average cortical neurons receive well over 1000 synapses (Beaulieu et al., 1992; Shepherd, 1998), and multiplying that by the number of neurons (but note that the most common type of neuron in the brain, the granule cells of the cerebellum, actually receive very few synapses—around 10). The estimate of 20 billion Web pages was based on the 2010 value from
http://www.worldwidewebsize.com
(the Google indicator). I consider the estimate of 1 trillion links to be an overestimate, which I have based on the average number of links per page times the total number of pages; estimates of the average number of links on a page (the out degree) are less than 10 (Boccalettii et al., 2006), but to ensure an over-rather than underestimate, I used a value of 50.
14
McGurk and MacDonald, 1976. There are many demos of this effect on the Web, including at
www.brainbugs.org
.
15
The notion that learning and cognition rely on associations between events and concepts that occur simultaneously or sequentially (contiguously) is an ancient one in philosophy and psychology. From Aristotle, through John Locke, James Mills, John Watson, and later Donald Hebb, and many “connectionist” modelers, the formation of associations is pivotal to classical and operant conditioning, language acquisition, and cognition in general. But as Steven Pinker has stressed, there is no doubt that there are other principles contributing to the generation and organization of human cognition (Pinker, 1997, 2002). Nevertheless, there is no controversy relating to the importance of associations in mental processes. In neuroscience the importance of associations is reinforced by the experimental fact that, as predicted by Donald Hebb and others, when two neurons are reliably activated in close temporal proximity, the synapse between these two neurons can be strengthened (see Chapter 1).
16
Plassmann et al., 2008.
17
Linden, 2007.
18
Richard Dawkins has referred to this as a “misfiring” (Dawkins, 2006).
19
Routtenberg and Kuznesof, 1967; Morrow et al., 1997.
CHAPTER 1: THE MEMORY WEB
1
Brownell and Gardner, 1988.
2
Answers from an undergraduate psychology class: zebra, 20; elephant, 12; dog, 9; giraffe, 6; lion, 6; cheetah, 3; horse, 3; tiger, 3; cat, 2; dolphin, 2; bear, 1; cow, 1; eel, 1; kangaroo, 1; komodo dragon, 1; panda, 1; rabbit, 1; “swimmy,” 1; whale, 1.
3
Purves et al., 2008.
4
Collins and Loftus, 1975; Anderson, 1983.
5
Watts and Strogatz, 1998; Mitchell, 2009.
6
Nelson et al., 1998.
7
Quiroga et al., 2005.
8
The strength of the links between the nodes may have two related neurobiological underpinnings: (1) the strength of the synapses between the neurons participating in each node and (2) overlap in the neurons participating in each node. That is, nodes of related concepts, such as “brain” and “mind,” may “share” many of their neurons. The more of these “shared” neurons, the stronger the “link” between the concepts or “nodes” (Hutchison, 2003).
9
Goelet et al., 1986; Buonomano and Merzenich, 1998; Martin et al., 2000; Malenka and Bear, 2004.
10
Babich et al., 1965; Rosenblatt et al., 1966.
11
Cajal, 1894. Kandel provides a wonderful historic account of the theories of learning and memory (Kandel, 2006).
12
Bliss and Lomo, 1973.
13
Hebb, 1949.
14
The demonstration that paired pre- and postsynaptic activity can elicit long-term potentiation is an example of “multiples” in science. The phenomenon was demonstrated more or less simultaneously in at least four different laboratories: Gustafsson and Wigstrom, 1986; Kelso et al., 1986; Larson and Lynch, 1986; Sastry et al., 1986.
15
Kandel et al., 2000; Malenka and Bear, 2004.
16
There really is no single Hebb’s rule, but rather a potpourri of related rules. For example, the precise temporal relationship between presynaptic and postsynaptic neurons is often important; specifically, synapses tend to get stronger if a presynaptic neuron fires before the postsynaptic neuron, but weaker if the events take place in the reverse order (Abbott and Nelson, 2000; Karmarkar et al., 2002).
17
As stated by the psychologist James McClelland: “consider what happens when a young child sees different bagels, each of which he sees in the context of some adult saying ‘bagel.’…Let’s assume that the sight of the bagel gives rise to a pattern of activation over one set of units, and the sound of the word gives rise to a pattern of activation over another series of units. After each learning experience of the sight of the bagel paired with the sound of its name, the connections between the visual nodes and the auditory nodes are incremented” (McClelland, 1985). Yet the mechanisms by which associations are formed must be much more complex, and are not fully understood. For example, it is likely that a critical component of learning is that some neurons are already connected more or less by chance. If they are coactive, these synapses survive and are strengthened, whereas the synapses between neurons that are not coactive are lost or pruned.
18
Vikis-Freibergs and Freibergs, 1976; Dagenbach et al., 1990; Clay et al., 2007.
19
Wiggs and Martin (1998), Grill-Spector et al. (2006), and Schacter et al. (2007) discuss some models of priming. One possibility is that priming may be a result of short-term changes in synaptic strength. In addition to the long-term changes in synaptic strength that underlie long-term memory, synapses can become stronger or weaker every time they are used. Depending on the synapses involved, these changes can last up to a few seconds (Zucker and Regehr, 2002). Under this hypothesis, the presentation of the word
bread
would activate a population of synapses, some of which would also be subsequently activated by the word
butter
, but as a result of short-term synaptic plasticity they would be stronger the second time around by facilitating, or priming, the activation of those neurons representing
butter
. It is possible that priming is a result of a specific form of short-term synaptic plasticity common to inhibitory synapses, known as paired-pulse depression. In this scenario neurons activated by the prime word would synapse onto local inhibitory neurons close the neurons representing the target word. In response to the prime these inhibitory neurons would fire; when the target was presented these same inhibitory neurons would be activated again, but their synapses would be weaker as a result of paired-pulse depression. The net result would be that the normal balance of excitation and inhibition present in neural circuits would be shifted toward excitation, facilitating the activation of the target neurons.
20
Brunel and Lavigne, 2009. Note this and other models need not rely on the notion that activity spreads from one node to related nodes, but rather that related representations have shared nodes, that is, that there is an overlap among the neurons representing related concepts.
21
Castel et al., 2007.
22
http://www.ismp.org/Tools/confuseddrugnames.pdf
, retrieved November 10, 2010.
23
Cohen and Burke, 1993; James, 2004.
24
You can take a variety of Implicit Association Tests at the Web site:
http://implicit.harvard.edu/implicit
. The results will inform you whether you have an implicit association bias but will not provide your reaction times.
25
Greenwald et al., 1998.
26
Nosek et al., 2009.
27
Galdi et al., 2008.
28
Bargh et al., 1996.
29
Williams and Bargh, 2008. Another study examined the stereotypical view that women are poorer at math than men and that Asians have quantitative skills superior to non-Asians. Two groups of Asian American women were asked to perform a math test. Before the test, one group filled out a questionnaire that focused primarily on their gender; the other group a questionnaire that focused on their Asian heritage. The group primed for awareness of gender performed more poorly than the group primed for awareness of race (Shih et al., 1999).
30
Jamieson, 1992.
CHAPTER 2: MEMORY UPGRADE NEEDED
1
Thompson-Cannino et al., 2009.
2
Before: the O. J. Simpson criminal trial ended in 1995 and the Atlanta Olympics were in 1996.
3
A. Lipta, “New trial for a mother who drowned 5 children,”
The New York Times
, January 7, 2005; “Woman not guilty in retrial in the deaths of her 5 children,”
The New York Times,
July 27, 2005.
4
Loftus et al., 1978; Loftus, 1996.
5
Ross et al., 1994.
6
Since it is unlikely that all possible groups of neurons representing specific nodes are initially connected with weak synapses, it is not known how we can form associations between any possible pair of concepts. But it seems that this process is initially facilitated by a brain structure that does not actually store our long-term memories, but is critical to their organization: the hippocampus (Hardt et al., 2010). Additionally, recent research has shown that neurons seem to always be exploring by continuously creating and withdrawing synapses. Some of these will prove useful and become permanent, and presumably the site of information storage (Yang et al., 2009; Roberts et al., 2010).