Watching an anesthetic block emotional memory

Here is an intriguing observation by Alkiri et al. We usually recall emotional pictures or events better than neutral ones. They found that low levels of the inhalation anesthetic sevoflurane could block this effect, and with PET imaging found a corresponding suppression of amygdala to hippocampal effective connectivity. Here is their abstract and one summary figure:

It is hypothesized that emotional arousal modulates long-term memory consolidation through the amygdala. Gaseous anesthetic agents are among the most potent drugs that cause temporary amnesia, yet the effects of inhalational anesthesia on human emotional memory processing remain unknown. To study this, two experiments were performed with the commonly used inhalational anesthetic sevoflurane. In experiment 1, volunteers responded to a series of emotional and neutral slides while under various subanesthetic doses of sevoflurane or placebo (no anesthesia). One week later, a mnemonic boost for emotionally arousing stimuli was evident in the placebo, 0.1%, and 0.2% sevoflurane groups, as measured with a recognition test. However, the mnemonic boost was absent in subjects who received 0.25% sevoflurane. Subsequently, in experiment 2, glucose PET assessed brain-state-related activity of subjects exposed to 0.25% sevoflurane. Structural equation modeling of the PET data revealed that 0.25% sevoflurane suppressed amygdala to hippocampal effective connectivity. The behavioral results show that 0.25% sevoflurane blocks emotional memory, and connectivity results demonstrate that this dose of sevoflurane suppresses the effective influence of the amygdala. Collectively, the findings support the hypothesis that the amygdala mediates memory modulation by demonstrating that suppressed amygdala effectiveness equates with a loss of emotional memory.

Figure - The cerebral metabolic effects of 0.25% sevoflurane are shown. (A) Representative high-resolution PET scans. (B) Absolute (mean ± SD) regional metabolic changes (white bars, placebo, no anesthesia; dark bars, 0.25% sevoflurane; marked with * for P less than 0.05). (C) Relative percent decreases of regional metabolism. (D) Regional SPM results of sevoflurane induced metabolic suppression (Upper, sagittal; Lower, axial). E shows the regional thalamic finding (brain center) on a colorized MRI. The SPM effects are significant at P less than 0.001, uncorrected; displayed at P less than 0.005, with a 500-voxel extent.

Languages shape the nuts and bolts of our perception.

The debate over whether language nudges the way we actually see the world is being resolved, and what has been the prevailing dogma - that basic parts of perception are too low-level, too hard-wired, too constrained by the constants of physics and physiology to be affected by language - is breaking down. Lera Boroditsky at Standford comments on this.

I used to think that languages and cultures shape the ways we think. I suspected they shaped they ways we reason and interpret information. But I didn't think languages could shape the nuts and bolts of perception, the way we actually see the world. That part of cognition seemed too low-level, too hard-wired, too constrained by the constants of physics and physiology to be affected by language.

Then studies started coming out claiming to find cross-linguistic differences in color memory. For example, it was shown that if your language makes a distinction between blue and green (as in English), then you're less likely to confuse a blue color chip for a green one in memory. In a study like this you would see a color chip, it would then be taken away, and then after a delay you would have to decide whether another color chip was identical to the one you saw or not.

Of course, showing that language plays a role in memory is different than showing that it plays a role in perception. Things often get confused in memory and it's not surprising that people may rely on information available in language as a second resort. But it doesn't mean that speakers of different languages actually see the colors differently as they are looking at them. I thought that if you made a task where people could see all the colors as they were making their decisions, then there wouldn't be any cross-linguistic differences.

I was so sure of the fact that language couldn't shape perception that I went ahead and designed a set of experiments to demonstrate this. In my lab we jokingly referred to this line of work as "Operation Perceptual Freedom." Our mission: to free perception from the corrupting influences of language.

We did one experiment after another, and each time to my surprise and annoyance, we found consistent cross-linguistic differences. They were there even when people could see all the colors at the same time when making their decisions. They were there even when people had to make objective perceptual judgments. They were there when no language was involved or necessary in the task at all. They were there when people had to reply very quickly. We just kept seeing them over and over again, and the only way to get the cross-linguistic differences to go away was to disrupt the language system. If we stopped people from being able to fluently access their language, then the cross-linguistic differences in perception went away.

I set out to show that language didn't affect perception, but I found exactly the opposite. It turns out that languages meddle in very low-level aspects of perception, and without our knowledge or consent shape the very nuts and bolts of how we see the world.

Watching waves of activity sweep across the brain

This is sensory physiology in the age of YouTube. Peterson and colleagues have used a voltage sensitive dye technique to watch the wave of first sensory area and then motor area excitation that is caused by a tiny deflection of a face whisker of a mouse:

Single brief whisker deflections evoked highly distributed depolarizing cortical sensory responses, which began in the primary somatosensory barrel cortex and subsequently excited the whisker motor cortex. The spread of sensory information to motor cortex was dynamically regulated by behavior and correlated with the generation of sensory-evoked whisker movement. Sensory processing in motor cortex may therefore contribute significantly to active tactile sensory perception.

The video shows the response when a mouses whisker touches an edge:

The movement of the C2 whisker was filmed with a high-speed camera at 500 Hz in an awake behaving mouse during an active touch sequence. Sensorimotor cortex was simultaneously imaged with VSD. At the time indicated by the vertical dotted line, the whisker contacts the object evoking a spreading sensorimotor response, first in S1 and subsequently in M1. The single trial imaging of cortical activity and the behavioral filming are matched frame-by-frame, synchronized through TTL pulses.

Consciousness papers for January

Most downloaded in January 2008 from the ASSC archives:

1. Koriat, A. (2006) Metacognition and Consciousness. In: Cambridge handbook
of consciousness. Cambridge University Press, New York, USA. 1359 downloads
from 27 countries. http://eprints.assc.caltech.edu/175/
2. Sagiv, Noam and Ward, Jamie (2006) Crossmodal interactions: lessons from
synesthesia. In: Visual Perception, Part 2. Progress in Brain Research,
Volume 155. 925 downloads from 19 countries.
http://eprints.assc.caltech.edu/224/
3. Robbins, Stephen E (2007) Time, Form and the Limits of Qualia. Journal of
Mind and Behavior, 28 (1). pp. 19-43. 760 downloads from 14 countries.
http://eprints.assc.caltech.edu/333/
4. Gomes, Gilberto (2005) Is consciousness epiphenomenal? Comment on Susan
Pockett. Journal of Consciousness Studies, 12 (12). pp. 77-79. 739 downloads
from 12 countries. http://eprints.assc.caltech.edu/160/
5. Seth, A.K. and Baars, B.J. (2005) Neural Darwinism and Consciousness.
Consciousness and Cognition, 14. pp. 140-168. 644 downloads from 18
countries. http://eprints.assc.caltech.edu/163/

Your amygdala and your blood pressure

I was intrigued by this article, because it shows what I suppose must be going on in my brain as I notice my blood pressure increasing when my stress system ramps up. (Being an introspective retired professor with sufficient time, I am increasingly noticing small changes in my breathing, heart rate, blood pressure, and relative sympathetic versus parasympathetic activation that accompany changes in context.)

In the article Gianaros et al. use the well-know Stroop color-word interference task to generate stress responses in a group of defined college students. They connect stressor processing with the brainstem cardiovascular control mechanisms regulating blood pressure. People with higher stressor-evoked blood pressure reactivity displayed more activation of the amygdala, especially in the dorsal part that contains the central nucleus. Individuals showing greater blood pressure reactivity also had a lower amygdala gray matter volume, which itself predicted greater amygdala activation. In addition, greater stressor- evoked blood pressure reactivity was correlated with stronger functional connectivity between the amygdala and the pons areas in the brainstem, which is critical for blood pressure control, as well the perigenual anterior cingulate cortex.

The data suggest that the amygdala and some of its projection areas play a role in mediating individual differences in autonomic stress responses and hence vulnerability to psychological stressors.

Figure - A. Greater mean arterial blood pressure (MAP) reactivity varied with greater amygdala activation to the incongruent condition. A, Clusters of the left and right amygdala where MAP reactivity varied with activation after covariate control for sex. Parametric maps are projected onto coronal (top) and axial (bottom) sections of a template derived from study participants. B, MAP reactivity (change from a resting baseline) is shown as a function of mean-centered and standardized amygdala BOLD activation values extracted from the peak voxels of the left (L; open circles, dashed line) and right (R; closed circles, solid line) amygdala clusters profiled in A.

Figure - Greater MAP reactivity varied with stronger positive amygdala-pons functional connectivity. A, Statistical parametric maps derived from an ROI regression analysis identifying pons areas where MAP reactivity varied as a function of connectivity with left (top) and right (bottom) amygdala seed regions. B, MAP reactivity is plotted as a function of amygdala-pons connectivity coefficients for the left (L; open circles, dashed line) and right (R; closed circles, solid line) amygdala.

Oliver Sachs on Migraines

An interesting article on Migraines by Oliver Sachs in the Op-Extra section of The New York Times, focusing on the geometric hallucinations they so often evoke.

...when I first saw photographs of the Alhambra, with its intricate geometric mosaics, I started to wonder whether what I had taken to be a personal experience and resonance might in fact be part of a larger whole, whether certain basic forms of geometric art, going back for tens of thousands of years, might also reflect the external expression of universal experiences. Migraine-like patterns, so to speak, are seen not only in Islamic art, but in classical and medieval motifs, in Zapotec architecture, in the bark paintings of Aboriginal artists in Australia, in Acoma pottery, in Swazi basketry — in virtually every culture. There seems to have been, throughout human history, a need to externalize, to make art from, these internal experiences, from the decorative motifs of prehistoric cave paintings to the psychedelic art of the 1960s. Do the arabesques in our own minds, built into our own brain organization, provide us with our first intimations of geometry, of formal beauty?

Whether or not this is the case, there is an increasing feeling among neuroscientists that self-organizing activity in vast populations of visual neurons is a prerequisite of visual perception — that this is how seeing begins. Spontaneous self-organization is not restricted to living systems — one may see it equally in the formation of snow crystals, in the roilings and eddies of turbulent water, in certain oscillating chemical reactions. Here, too, self-organization can produce geometries and patterns in space and time, very similar to what one may see in a migraine aura. In this sense, the geometrical hallucinations of migraine allow us to experience in ourselves not only a universal of neural functioning, but a universal of nature itself.

Video game addiction, and taking play seriously.

Video games trigger reward and addiction centers of the brain, just like cocaine. Hoeft et. al. at Stanford have compared the activation of these centers in men's and women's brains as they played video games and found them to be more active in male than in female participants. Males showed greater activation and functional connectivity compared to females in the mesocorticolimbic system. The more the men won, the stronger their brain activity. Women's responses were less intense and didn't correlate with winning. This may have something to do with why men seem to become addicted to video games much more easily than women.

There is an interesting and more general article by Robin Henig on play in animals and humans in the New York Times Magazine. The general consensus is that play activity is very important in the development of social intelligence and the ability to respond to rapidly changing situations. It turns out that the rise and decay of play activity in animals corresponds closely to the development curve of the cerebellum, which is important in skilled movements. In one interesting study, experimenters:

...raised 12 female rats from the time they were weaned until puberty under one of two conditions. In the control group, each rat was caged with three other female juveniles. In the experimental group, each rat was caged with three female adults. Pellis knew from previous studies that the rats caged with adults would not play, since adult rats rarely play with juveniles, even their own offspring. They would get all the other normal social experiences the control rats had — grooming, nuzzling, touching, sniffing — but they would not get play... (in) the rats raised in a play-deprived environment, they found a more immature pattern of neuronal connections in the medial prefrontal cortex... less selective pruning of cells and a more tangled, immature medial prefrontal cortex in play-deprived rats might mean that the rat will be less able to make subtle adjustments to the social world.

There are numerous theories about the function of play. It doesn't seem unreasonable that that the fragmentary, disorderly, unpredictable, exaggerated, improvised, vertiginous, and nonsensical nature of play trains the brain allows for a wider behavioral repertory and perhaps more competence in responding to novel or unforeseen situations.

Killer Instincts.

Dan Jones writes a news feature in Nature on neuroscientific and evolutionary perspectives on homicide, mainly carried out by men. Here are some selected chunks:

Men are not just more likely to kill other people than women are, they are also more likely to do so in groups ...Humans are not the only primates to form coalitions that kill members of neighbouring communities. ...five long-term study sites dotted around Africa have seen murderous 'gang violence' in chimpanzees...Wilson and Muller have compared death rates from conflict between groups of chimps in the five long-term study sites with data for inter-group human conflicts in numerous subsistence-farmer and hunter–gatherer societies...Overall, humans and chimpanzees showed comparable levels of violent death from aggression between groups...however, chimps display within group aggression and killing behaviour 200 times more frequently that aboriginal human groups...this prosocial lack of violence looks like a fundamental aspect of human nature — the human ability to generate in-group amity often goes hand in hand with out-group enmity...Choi and Bowles have produced models in which altruism and war co-evolve, promoting conflict between groups and greater harmony within them.

A decline in inter-personal violence (as opposed to inter-group war) can be seen over the shorter timescale and narrower field of modern European history. Eisner has documented a trend of declining homicide rates estimated from historical records left by coroners, royal courts and other official sources spanning Europe from the twelfth century to the modern day. After rising from an average of 32 homicides per 100,000 people per year in the thirteenth and fourteenth centuries to 41 in the fifteenth, the murder rate has steadily dropped in every subsequent century, to 19, 11, 3.2, 2.6 and finally 1.4 in the twentieth century...a few centuries is too short a time for evolution to have shaped human nature much... A part of the answer that is consistent with an evolutionary approach is a long-term reduction in inequalities of life circumstances and prospects.

Human and Animal Math

Michael Beran writes a brief review of the evolutionary and developmental foundations of mathematics. Humans and other higher animals are born with a dedicated systems for numerical processing.

The view from my window....

The title line for this posting is stolen from Andrew Sullivan's blog -

This is the view looking out of the window by the piano in my Wisconsin home yesterday (the reason I am in Ft. Lauderdale right now), and below is the outside view of that window this past spring.

T

Persuasion, bonding, and social mimicry

Benedict Cary mentions some interesting studies on social mimicry, persuasion, and affiliative behavior in an article from the NY Times science section. (Chapter 5 of my book has some stuff on this topic.) In one experiment he cites, the experimenter:

...had student participants go to a lab and give their opinions about a series of advertisements. A member of his research team mimicked half the participants while they spoke, roughly mirroring the posture and the position of their arms and legs, taking care not to be too obvious. Minutes later, the experimenter dropped six pens on the floor, making it look like an accident. In several versions of this simple sequence, participants who had been mimicked were two to three times as likely to pick up the pens as those who had not. The mimicry had not only increased good will toward the researcher within minutes, the study concluded, but it also prompted “an increased pro-social orientation in general.”

Another experiment tested how being mimicked might affect the behavior of a potential client or investor:

The team had 37 Duke students try out what was described as a new sports drink, Vigor, and answer a few questions about it. The interviewer mimicked about half the participants using a technique of mirroring a person’s posture and movements, with a one- to two-second delay. The idea is to be a mirror but a slow, imperfect one. Follow too closely, and most people catch it — and the game is over. None of the copied participants picked up on the mimicry. But by the end of the short interview, they were significantly more likely than the others to consume the new drink, to say they would buy it and to predict its success in the market. In a similar experiment, the psychologists found that this was especially true if the participants knew that the interviewer, the mimic, had a stake in the product’s success.

The article gives several other example of the subconscious social waltz, or kinesic communication, that underlies smooth human communication.

Need something to worry about? Climate tipping points...

This graphic is from an open access article in PNAS by Lenton et al. on tipping elements in the earth's climate system. You probably need to click on the graphic to make it larger; the color indicating the population densities is hard to see. In the same issue of PNAS, there is an article on how when it get warm (as in the Paleocene – Eocene Thermal Maximum caused by a carbon dioxide increase about 55 million years ago), the insects chow down on the plants.

Legend for graphic - Map of potential policy-relevant tipping elements in the climate system, overlain on global population density. Subsystems indicated could exhibit threshold-type behavior in response to anthropogenic climate forcing, where a small perturbation at a critical point qualitatively alters the future fate of the system. They could be triggered this century and would undergo a qualitative change within this millennium. We exclude from the map systems in which any threshold appears inaccessible this century (e.g., East Antarctic Ice Sheet) or the qualitative change would appear beyond this millennium (e.g., marine methane hydrates). Question marks indicate systems whose status as tipping elements is particularly uncertain.

High-Functioning Autism: a neural phenotype in the cingulate cortex

A review by Chris and Uta Frith discusses and important paper in Neuron from Montague's group in Houston, who:

...have measured brain activity (using fMRI) while volunteers, who are classified as being at the high-functioning end of the autistic spectrum, were engaged in a simple social interaction. The task was an iterated trust game in which two subjects take turns as investor or trustee. The investor chooses how much to money to invest. This chosen amount is tripled on its way to the trustee, and the trustee then chooses how much to repay to the investor. Read Montague and his colleagues have studied this game extensively in large groups of volunteers and have observed a characteristic pattern of brain activity in the anterior cingulate cortex. When making an investment (self phase), transient increases in activity are seen in an area of mid cingulate cortex (−7 <>A graphic from the Chiu et al paper showing the diminished "self" response in autism spectrum patients:


...the results suggest that the abnormality associated with autism is restricted to only one phase of the interactive game: the point where the autistic volunteer makes an investment, not the point where the autistic volunteer is told about the repayment made by their partner. Additional results from Read Montague's group give further clues as to the implications of this result. First, the same pattern of activity in cingulate cortex is observed when volunteers are shown pictures of people engaged in athletic activities and asked to imagine themselves taking part. This is further evidence as to the nature of the cognitive process associated with this pattern of activity: it involves thinking about the self acting in a social context. Second, the characteristic patterns of activity in the cingulate cortex are only observed when the trust game is played with a human partner. No such distinct patterns emerge when the game is played in the absence of a responsive social partner...At least part of the imagining must involve thinking about how one would fit in with the group, and how other group members would evaluate one's performance. Actually, this is a question about the kind of reputation one might gain in the eyes of the others. Likewise, in the self phase of the trust game, the amount one invests can be seen as a measure of how much one trusts one's partner. It is not just giving an amount of money; it is giving a signal to the other person: “trust me” and “I trust you.”
In other words, at the point of investment we are predicting what the effect of our investments is going to be on the behavior of our partners. In the other phase of the game, we are also evaluating a signal. But there is a difference. The evaluation is after the fact. We know what the investment is. We are not at this point trying to build our reputation in the other player's eyes.

A non-invasive brain-machine interface?

Waldert et al. show that hand movement direction can be decoded from magneto- (MEG) and electro-(EEG) encephalography recordings. They suggest that this might make it possible to design a non-invasive brain machine interface. Here is their abstract:

Brain activity can be used as a control signal for brain–machine interfaces (BMIs). A powerful and widely acknowledged BMI approach, so far only applied in invasive recording techniques, uses neuronal signals related to limb movements for equivalent, multidimensional control of an external effector. Here, we investigated whether this approach is also applicable for noninvasive recording techniques. To this end, we recorded whole-head MEG during center-out movements with the hand and found significant power modulation of MEG activity between rest and movement in three frequency bands: an increase for ≤7 Hz (low-frequency band) and 62–87 Hz (high-{gamma} band) and a decrease for 10–30 Hz (β band) during movement. Movement directions could be inferred on a single-trial basis from the low-pass filtered MEG activity as well as from power modulations in the low-frequency band, but not from the β and high-{gamma} bands. Using sensors above the motor area, we obtained a surprisingly high decoding accuracy of 67% on average across subjects. Decoding accuracy started to rise significantly above chance level before movement onset. Based on simultaneous MEG and EEG recordings, we show that the inference of movement direction works equally well for both recording techniques. In summary, our results show that neuronal activity associated with different movements of the same effector can be distinguished by means of noninvasive recordings and might, thus, be used to drive a noninvasive BMI.

Creating new worlds - science and fiction as the same thing

Alison Gopnik has an interesting take on why children pretend much of the time and spend time in imaginary worlds:

Recently, I've had to change my mind about the very nature of knowledge because of an obvious, but extremely weird fact about children - they pretend all the time. Walk into any preschool and you'll be surrounded by small princesses and superheroes in overalls - three-year-olds literally spend more waking hours in imaginary worlds than in the real one. Why? Learning about the real world has obvious evolutionary advantages and kids do it better than anyone else. But why spend so much time thinking about wildly, flagrantly unreal worlds? The mystery about pretend play is connected to a mystery about adult humans - especially vivid for an English professor's daughter like me. Why do we love obviously false plays and novels and movies?

The greatest success of cognitive science has been our account of the visual system. There's a world out there sending information to our eyes, and our brains are beautifully designed to recover the nature of that world from that information. I've always thought that science, and children's learning, worked the same way. Fundamental capacities for causal inference and learning let scientists, and children, get an accurate picture of the world around them - a theory. Cognition was the way we got the world into our minds.

But fiction doesn't fit that picture - its easy to see why we want the truth but why do we work so hard telling lies? I thought that kids' pretend play, and grown-up fiction, must be a sort of spandrel, a side-effect of some other more functional ability. I said as much in a review in Science and got floods of e-mail back from distinguished novel-reading scientists. They were all sure fiction was a Good Thing - me too, of course, - but didn't seem any closer than I was to figuring out why.

So the anomaly of pretend play has been bugging me all this time. But finally, trying to figure it out has made me change my mind about the very nature of cognition itself.

I still think that we're designed to find out about the world, but that's not our most important gift. For human beings the really important evolutionary advantage is our ability to create new worlds. Look around the room you're sitting in. Every object in that room - the right angle table, the book, the paper, the computer screen, the ceramic cup was once imaginary. Not a thing in the room existed in the pleistocene. Every one of them started out as an imaginary fantasy in someone's mind. And that's even more true of people - all the things I am, a scientist, a philosopher, an atheist, a feminist, all those kinds of people started out as imaginary ideas too. I'm not making some relativist post-modern point here, right now the computer and the cup and the scientist and the feminist are as real as anything can be. But that's just what our human minds do best - take the imaginary and make it real. I think now that cognition is also a way we impose our minds on the world.

In fact, I think now that the two abilities - finding the truth about the world and creating new worlds-are two sides of the same coins. Theories, in science or childhood, don't just tell us what's true - they tell us what's possible, and they tell us how to get to those possibilities from where we are now. When children learn and when they pretend they use their knowledge of the world to create new possibilities. So do we whether we are doing science or writing novels. I don't think anymore that Science and Fiction are just both Good Things that complement each other. I think they are, quite literally, the same thing.

The Best Men Are (Not Always) Already Taken

Bressan1 and Stranieri take a (dubious) evolutionary psychological approach to the question of female preference for single versus attached males. The outcome is interesting. Here is their abstract:

Because men of higher genetic quality tend to be poorer partners and parents than men of lower genetic quality, women may profit from securing a stable investment from the latter, while obtaining good genes via extrapair mating with the former. Only if conception occurs, however, do the evolutionary benefits of such a strategy overcome its costs. Accordingly, we predicted that (a) partnered women should prefer attached men, because such men are more likely than single men to have pair-bonding qualities, and hence to be good replacement partners, and (b) this inclination should reverse when fertility rises, because attached men are less available for impromptu sex than single men. In this study, 208 women rated the attractiveness of men described as single or attached. As predicted, partnered women favored attached men at the low-fertility phases of the menstrual cycle, but preferred single men (if masculine, i.e., advertising good genetic quality) when conception risk was high.

Painful or disgusting? - ask parietal and cingulate cortex

Benuzzi et al. find brain activity specific for disgusting scenes in the posterior cingulate cortex. Signal changes in perigenual cingulate and left anterior insula are linearly related to the perceived unpleasantness. Painful scenes selectively induce activation of left parietal foci including the parietal operculum, the postcentral gyrus, and adjacent portions of the posterior parietal cortex. Their abstract is here, and here is one figure from the paper:

Top, Cortical foci active during the observation of painful video clips. Bottom, Cortical foci active during the observation of disgusting video clips.

Are today's young people really more narcissistic?

A comment by Trzesniewski et al. on the idea that today's young people have inflated impressions of themselves compared with previous generations. They:

..investigated secular trends in narcissism and self-enhancement over the past three decades. Despite recent claims about the impact of the "self-esteem movement" on the current generation of young people, we found no evidence that college students' scores on the Narcissistic Personality Inventory increased from the 1980s through 2007 (N = 26,867), although we did find small changes in specific facets of narcissism. Similarly, we found no evidence that high school students' level of self-enhancement, defined by the discrepancy between their perceived intelligence and their actual academic achievements, increased from 1976 to 2006 (N = 410,527). These results cast doubt on the belief that today's young people have increasingly inflated impressions of themselves compared with previous generations.

100% accuracy in automatic face detection.

A problem with the automatic face recognition systems being tested in some airport security screening systems is that none can cope with the kind of image variability encountered in the real world. Jenkins and Burton have used a simple averaging technique to increase the accuracy of an industry standard face-recognition algorithm from 54% to 100%. They averaged the images from 20 different photographs for each of 25 male celebrities who were also in a large public online database of 31,077 photographs of famous faces, comprising an average of nine different photos for each of 3628 celebrities - these images were highly variable in their quality and covered a wide range of lighting conditions, facial expressions, poses, and age. Using the FaceVACS (Cognitec Systems GmbH, Dresden, Germany)industry standard face-recognition system that has been widely adopted, they fed this database their averaged images for each of 25 male celebrities who were also in the online database (excluding photos that were both in their sample and in the database). With the averaged images, the database returned the correct identification 100% of the time. When individual photographs were presented to the database the correct identification was returned only ~50% of the time. From their text:

We demonstrated this improvement with a commercially available algorithm and an online face database over which we had no control. We suggest that image averaging enhances performance by stabilizing the face image. With standard photographs, the match tends to be dominated by aspects of the image that are not diagnostic of identity (e.g., lighting and pose). Averaging together multiple photographs of the same person dilutes these transients while preserving aspects of the image that are consistent across photos. The resulting images capture the visual essence of an individual's face and elevate machine performance to the standard of familiar face recognition in humans. It would be technically straightforward to incorporate an average image into identification documents. Doing so would greatly reduce the incidence of face-recognition errors and raise the prospect of a viable automatic face-recognition infrastructure.


Example photographs of Bill Clinton and their average (right). [Image 1, photo by Marc Nozell (www.flickr.com/photos/marcn/534512066); image 2, photo by Roger Goun (www.flickr.com/photos/sskennel/829574139); image 20, photo by Nelson Pavlosky (www.flickr.com/photos/skyfaller/26752190). All photos were used under a Creative Commons license.] Different pictures of a single face can vary enormously, making automatic recognition difficult. Averaging together multiple photos of the same face stabilizes the image, improving performance dramatically.

Need an antidepressant?