The Electroencephalogram Cocktail Party

In 2005, filmmaker Joyce Draganosky made a short entitled The Science of Love, where...

...the battle between reason and emotion takes center stage. A professor, who believes she has found a way of determining scientifically whether someone is in love, clashes with her department chair, a woman who thinks love and attraction are far too complex to be mapped according to the certainties of science.

The clip above highlights a hilarious event designed to determine the neural correlates of love. How does it work?

"Well, the goal is to identify the part of the brain that controls lust, and to show that it is different from the part of the brain that controls love. In order to do this we had to simulate as natural an environment as possible while still being able to measure neural activity. So I have created the Electroencephalogram Cocktail Party. An EEG Mixer!!"

Draganosky received an Alfred P. Sloan Foundation grant for The Science of Love. She spent a year researching the film, attending presentations and conducting interviews in the laboratory of Dr. Joy Hirsh at Columbia. "I actually audio-taped the interviews because I wanted to get all the scientist-speak perfectly right so that I could write accurate dialogue." We can certainly forgive the confusion of fMRI and EEG results for artistic sake (and comic effect) here. The lead character Syd seems to be modeled after Dr. Helen Fisher, even down to the combination of evolutionary anthropology with brain imaging.¹


The Real EEG Mixer

Now, an actual study by Gevins and colleagues (2012) recorded EEG while 10 members of his lab attended a cocktail party furnished with food and alcohol:

The cocktail party was unscripted, other than withholding drinking for the first 10 minutes to record a pre-drinking baseline. The partiers intermingled, chatted, ate sushi and hors d'oeuvres and drank vodka martinis or vodka and cranberry cocktails according to their personal inclinations. They also measured BACs [breath alcohol content], took photos and checked up on the automated data collection.

The goal here was not to evaluate the effectiveness of EEG Speed Dating or to explicate a lofty Science of Love, but instead to examine the effects of alcohol on the spectral properties, or frequency composition, of EEG in a naturalistic setting. Although the technique of "hyperscanning"³ has been used in fMRI studies, the advantages of EEG recording in this context should be obvious.⁴

The participants served as both subjects and experimenters:

Teams placed headsets simultaneously on groups of participants; total set up time was about a half hour. Data from each EEG headset was transmitted during the party via Bluetooth protocol to its own dedicated notebook computer. The data were time synchronized across computers by a start signal sent via local Ethernet from one computer to all other recording computers.

They measured breath alcohol contents, took photos and checked the data collection.⁵



Using data obtained from two prior experiments (collected from separate groups of subjects in a more typical recording environment), multivariate divergence analyses searched for the subset of variables⁶ that best discriminated between the EEGs of brains on alcohol vs. brains on placebo. These equations were applied to the pre- and post-alcohol EEGs of the partygoers. Although 40% of the ambulatory cocktail party data were lost due to artifacts, the remaining data showed 80% sensitivity in recognizing alcohol and 80% specificity in recognizing no-alcohol. While mandatory Wearable Wireless EEG Fashion Accessories will not be replacing breathalyzers at sobriety stops any time soon, this was certainly a fun pilot study (especially for the participants)!


Footnotes

¹ Recycling Alert: this paragraph contains portions of my original 2006 review of the entire 18 minute short (combined with additional information). It's only two sentences, but I need to be forthcoming in anticipation of my future blogging gigs at Wired and the New Yorker.

² It looks like Neuroskeptic has already posted coverage of this study: Cocktail-Party Neuroscience.

³ The hyperscanning technique was developed by Read Montague and co-workers (Montague et al., 2002). Typically, hyperscanning involves two subjects who interact with each other while playing a "trust" game. Their brains are scanned simultaneously (in adjoining or distant magnets). Specially developed software coordinates the experiments across sites.

⁴ You can't walk around, eat, drink, and directly interact with others while lying in a scanner.

⁵ Who's real and who's fictional??

⁶ Technical details:

The individual frequency component powers were then averaged into 3 standard bands, theta (4–7 Hz), alpha (8–13 Hz) and beta (13–18 Hz) determined in prior studies to be sensitive to alcohol's effect on the EEG. The 3 banded powers and their standard deviations constituted a total set of 6 variables for each of the 7 electrode sites.

This is a very low number of variables; compare to a recent structural MRI study that used 231 neuroanatomical variables to predict brain age.


Reference

Gevins A, Chan CS, & Sam-Vargas L (2012). Towards measuring brain function on groups of people in the real world. PloS one, 7 (9) PMID: 22957099