"The Brighthouse Institute for Thought Sciences

December 4, 2002 8:32 AM

"The Brighthouse Institute for Thought Sciences plans to change the marketing world by using modern neuroscience methods to observe and understand the true drivers of consumer behavior. The Thought Sciences team uses functional Magnetic Resonance Imaging (fMRI), a safe and non-invasive technique, to observe patterns in brain activity that reveal how a person is processing and/or evaluating a product, object or advertisement." (From their press release.) Is this the next logical step beyond focus groups? And does this seem just a little bit creepy to anyone else?
posted by Johnny Assay (25 comments total)

MRI? Pfui, that's for amatuers. What you need's a SQUID.
posted by adamgreenfield at 8:38 AM on December 4, 2002

All the science fiction I read as a boy in the early sixties is coming true. At least the dystopian stuff.
posted by kozad at 8:57 AM on December 4, 2002

How is one person's reaction will be relevant to everyone else? Are we that similar in our thinking and neurological reactions? Not if you get your clues from reading MeFi.
posted by semmi at 8:58 AM on December 4, 2002

Nice try semmi, but I am sure they use the regular technique of gathering samples across a group and then extrapolating predictions with some sort of statically supported math. Individuals are fairly difficult to predict unless you know them very well, but larger groups of people are easier.

Further still think of the possibilities for individual manipulation. Imagine a point in the future (never quite as far away as it used to seem) where the scanning technology has become very cheap to produce and can scan at a distance of several meters. Far fetched I realize. Imagine also that, subway ads for instance, are not printed but displayed on liquid crystal panels (less far fetched, in fact, seems like a good idea). The scanner registers you as you approach, measuring your responses to its display, all the while using the feedback to subtly modulate its display to optimize your reactions according to certain pre-programmed desired response patterns. Neurologically customized advertisements. Further still, why scan every time, build a profile, associate it with a brainprint, share it across the network...
posted by randomnfactor at 9:16 AM on December 4, 2002

I've read a little bit about fMRI studies used for language processing. I really know very little about them, but I think I know enough to speculate that these people won't have any easier of a time finding conclusive answers to how the brain processes things than anyone else working with fMRI or any other brain analysis devices. In fact, what they want looks much harder.

It's simply too complex to point your brain scan toy at the black box that is our brains and get answers - you need many rigorous, controlled, and difficult experiments run by experienced experimenters in order to get believable results.

Not only that, most tasks that people do brain imagining/ERP/whatever studies on are much much simpler than "evaluating a product".
For at least the things I'm familiar with, they are tasks where there is good reason to suspect that processing is handled by a somewhat specific part of the brain, and great effort is taken to eliminate possible interference from other mental tasks. I'm not sure that this will be very easy to do for what they want.

on preview: randomnfactor: most tasks that people deal with in neurological studies are tasks where there is reason to believe people will generally all do the same thing in a somewhat predictable way. It's not clear that their task is anything like this at all, and it seems like this would be necessary for statistical analysis.
posted by advil at 9:27 AM on December 4, 2002

random(n):

Imagine also that, subway ads for instance, are not printed but displayed on liquid crystal panels (less far fetched, in fact, seems like a good idea). The scanner registers you as you approach, measuring your responses to its display, all the while using the feedback to subtly modulate its display to optimize your reactions according to certain pre-programmed desired response patterns.

this won't happen. not with subway ads, at any rate: the possibility for damage is too great. LCDs are expensive, and shall likely always be more expensive than print ads, and all you have to do to break an LCD is throw something heavy at it. billboard ads, in some areas of some cities, may have better luck, but i'm sure those would be many times more expensive due to their larger sizes.

Further still, why scan every time, build a profile, associate it with a brainprint, share it across the network...

i think the problem with proposed futuristic marketing techniques is that they're way too expensive. holy moley: networked ad boards? transferring something as complex as human response? i have a hard time believing this will be the future as opposed to the far cheaper routes people go now.
posted by moz at 9:33 AM on December 4, 2002

fMRI scanners don't read minds. They measure the flow of blood in the brain and only that. Snake oil scientists are hyping the scans and their pretty pictures as much more but there are all kinds of largely unsupported assumptions being made by researchers about what the blood flow means. Could you predict where a car would go based on the flow of gasoline in part of the engine? A lot people think that brain imaging is the new phrenology and I tend to agree.
posted by srboisvert at 10:25 AM on December 4, 2002

Here's an article on the Brighthouse Institute, via /.

Neuromarketing sounds like the dark side of the kind of work Andrew Newberg does, studying the brains of meditating Buddhist monks.
posted by homunculus at 10:49 AM on December 4, 2002

Similar previous post.
posted by homunculus at 11:01 AM on December 4, 2002

"i think the problem with proposed futuristic marketing techniques is that they're way too expensive."

maybe your imagination is just too limited. i don't buy any immediate plans to read intentions through technology, but the idea of networked displays seems very immediate. most of those airport displays and kiosks you see in malls are ALREADY networked.
posted by muppetboy at 11:16 AM on December 4, 2002

if you want to see the very near future, look at this. Wireless peer-to-peer networking technology could be made very cheap for millions or even billions of e-ink billboards, posters, papers etc. In sufficient quantities, e-ink may be almost as cheap as paper. When you consider that the message can vary to display the messages that would take reams of paper to display statically, it may in the end be considerably cheaper.
posted by muppetboy at 11:19 AM on December 4, 2002

Advil writes: "you need many rigorous, controlled, and difficult experiments run by experienced experimenters in order to get believable results."
So true, but this being the marketing world, it's not believable results BrightHouse needs, just saleable results. To follow srboisvert's line of reasoning, the snake oil need not be of any particular quality. Gotta love the irony...

This article in Salon.com is pretty interesting, too.
posted by micropublishery at 11:24 AM on December 4, 2002

The heart of the matter from the Salon article I linked to above:

"Just as early mariners were clueless about the Coriolis forces that drove the trade winds, so have advertisers honed their craft without understanding the neural underpinnings of the desires and perceptions on which they seek to capitalize. And just as understanding the physics of trade winds had little practical effect on maritime navigation, so, too, will the march of cognitive science be unlikely to spin off a scientific field of neuroadvertising.

"For about 50 years there have been people who looked at physiological or neurological responses to marketing stimuli," says David Stewart, a marketing expert at the University of Southern California. "[But] they have not had much real impact on the way advertising is created or assessed. The reality is that it's almost always easier to just directly ask people what they liked about an ad," or follow their behavior afterward to gauge its effectiveness.

From this point of view, neuroscience will probably not soon be adding any tools to the marketing industry's bag of tricks. But as these examples show, it will at least help explain how the current tools work. Indeed, as cognitive science emerges from its own Bronze Age, it is holding up the hope (or for some onlookers, the horror) of a new set of concepts for charting the lay of human nature."
posted by micropublishery at 11:47 AM on December 4, 2002

muppet:

maybe your imagination is just too limited. i don't buy any immediate plans to read intentions through technology, but the idea of networked displays seems very immediate. most of those airport displays and kiosks you see in malls are ALREADY networked.

maybe my imagination really is just too limited. does it matter?

i don't think airport displays and mall kiosks are good analogies for subway ads and billboards. it may be costly, but not very difficult, to get kiosks and airport displays to work together via networking. moreover, since those technologies are contained wholly within the airport or the mall area, they're simpler to get at than many billboards or MT ads. disparate locations increase costs.

Wireless peer-to-peer networking technology could be made very cheap for millions or even billions of e-ink billboards, posters, papers etc. ... When you consider that the message can vary to display the messages that would take reams of paper to display statically, it may in the end be considerably cheaper.

the cost of enabling wireless P2P networking might not be too bad in the future, but the true cost of networking is in maintenance. if someone sprays graffiti onto a train ad, it's not such a bad thing; if the network goes down, it's a pretty bad thing as all the ads will cease to function for a train stop. certainly the skill of labor required to perform such maintenance will be higher than it is to clean off or replace print ads, and as such the cost of that labor will also rise.
posted by moz at 11:52 AM on December 4, 2002

moz, with enough ads and a simple, resilient peer-to-peer design, the network will generally not go down. in the fairly near future (5 years, maybe), there will be no more skill involved in pasting up e-ink-like ads than there will be in pasting up paper. the ads can hook themselves into the network, authenticate, download ad info, etc. automatically. when they malfunction, you just paste up another e-ad sheet (probably over top of the old one just like paper ads). ubiquitous paper-cost e-ads are probably more than 5 years out, but i'd bet not more than 10. i think it's going to be a *very* loud, annoying, blinking world in 10 years.

here's something to think about in terms of technology costs. at one point chips were pretty expensive. but in a device i'm working on, the chip we're using costs $0.60 while the black-and-white sticker we're putting on the back costs $1.20! my bet is that simple, low-bandwidth wireless networking technology goes to $0.60 or below in just a few years. e-ink may take longer to get that cheap or it may not. depends on innovation and competition more than anything technical.
posted by muppetboy at 1:14 PM on December 4, 2002

muppetboy:

moz, with enough ads and a simple, resilient peer-to-peer design, the network will generally not go down.

a large number of ads, or rather network nodes, should have no positive effect on network consistency. that aside, "resilient peer-to-peer design" is an interesting choice of words.

if all networks were "resilient," they would generally not go down; that's a truism of networking, but by that token it does not offer much to our discussion. if all network nodes were in communication with each other, which i infer is what you mean when you write "peer-to-peer," i would not consider that a mark of especial resilience either; rather a feature that i would regard necessary in a distributed network. (ethernet is a great example.)

wireless networking is something that i am wary of. can i spoof a network node, and tell the billboards to display pictures from goatse.cx? my current impression is that a lot of security would need to go into maintaining a wireless network. would the cost of that really be less than a couple of people pasting up print ads, or would the utility of changing specific ads at will really be worth it?

again, the skill of those pasting up ads is not something i am interested in. it's the skill involved in maintaining the network as a whole. but then again, perhaps we'll that the benefits outweigh the costs?
posted by moz at 2:18 PM on December 4, 2002

micropublishery: it's not believable results BrightHouse needs, just saleable results.

Indeed. Note that all three of BrightHouse's principals are marketers, not scientists.

You can imagine their "brainstorming" sessions to name the company: "It's got to be solid, like a building, but illuminating..." "How about SunBarn?" "FlashCastle?" "ShinyBrick?" "BrightHouse!" High fives all around.

What a stupid idea. Their clients deserved to be fleeced.
posted by skimble at 2:38 PM on December 4, 2002

As someone who paid for his previous life delinquencies in the form of a penitence generally referred to as 'retail sales', I recall on many occasions the odd occurrence of widget X being sought after by a hoard of people who, in the course of one night, 'had to have it'. It could have sat on a shelf collecting dust for years, only to be sold out in the time span of twenty minutes.

..Or the similar stories I've heard from food service employees who, during the course of one night, had a run on an odd particular item, one that may have been normally shunned and overlooked.

Call it Jungian phenomena, mere coincidence, or whatever.

That being said, I would think that statisticians would have the same sort of luck running probabilities on something as diverse as consumer spending habits as they would on any complex system (like the weather for instance).

And using any sort of brain imaging as a predictor of large scale behavior? I understand the concept of sampling, but this sounds like looking at the books of a particular business as a means to form predictions of the stock market... Or using measurements of quantum level interactions as a means to speculate on cell level biology.

However, maybe I am missing the point.

BTW, moz, your mini discussion seems very similar to the sort of discussions that monks and scribes may have had about widespread literacy over a good parchment scraping before the invention of the printing press.. Many futurists seem to make the mistake of forecasting the future while operating under the technological constraints of the present.
posted by jazzkat11 at 6:20 PM on December 4, 2002

actually, on second thought, I think the subway ads of the future will read something like:

C'mon now Jim. Buy the red stapler. We know you want to.
This message brought to you by the scanning of blood flow into your collectivist meme-dula

(ok, well that was at least humorous to this sleep deprived mind)
posted by jazzkat11 at 6:46 PM on December 4, 2002

Many futurists seem to make the mistake of forecasting the future while operating under the technological constraints of the present.

Seems to me that most futurists over-lard the future by underestimating technological/economical constraints.
posted by Opus Dark at 8:03 PM on December 4, 2002

"a large number of ads, or rather network nodes, should have no positive effect on network consistency. that aside, "resilient peer-to-peer design" is an interesting choice of words."

>>> the number of nodes in a peer-to-peer network has a very direct impact on the resiliency of the whole. the internet itself is fairly resilient because it has, to some extent, a design like this. in a peer-to-peer network where each node is interchangable with any other, local problems can be routed around.

essentially it works like this: the more nodes in your network and the more distributed (non-hierarchical) the network is, the more ways there are to route around failures. if you reduce the number of nodes or change the topology to introduce hierarchical dependencies you will create inequalities between the nodes and therefore inflexibilities which will increase the possibility of systemic failure.

"if all networks were "resilient," they would generally not go down; that's a truism of networking, but by that token it does not offer much to our discussion. if all network nodes were in communication with each other, which i infer is what you mean when you write "peer-to-peer," i would not consider that a mark of especial resilience either; rather a feature that i would regard necessary in a distributed network. (ethernet is a great example.)"

>>> it is impossible to create a network of any real scale with nodes in direct communication with each other because the combinatorics explode too quickly. peer to peer networking is resilient specifically because nodes are only incommunication with handfuls of "nearby" nodes. generally, the number of nodes is "enough" to more or less eliminate failure points. with a good topology, a peer to peer network can route around failures better than any other system i know of.

"wireless networking is something that i am wary of. can i spoof a network node, and tell the billboards to display pictures from goatse.cx? my current impression is that a lot of security would need to go into maintaining a wireless network. would the cost of that really be less than a couple of people pasting up print ads, or would the utility of changing specific ads at will really be worth it?""

>>> cryptography and digital signatures should generally be able to solve this problem. the main cost to the digital e-ink ad would be a slightly higher powered cpu if you wanted the poster to update instantly. if you can wait a few seconds or perhaps minutes, you could probably get the job done with that $0.60 cpu i was describing.

>>> maybe i've got this all wrong, but i still don't see any serious technological or economic barrier to pervasive e-ink ads within 10 years. a company like e-ink will make billions off this if they price their technology right. if they don't, there are other similar plans at other companies.
posted by muppetboy at 8:51 PM on December 4, 2002

essentially it works like this: the more nodes in your network and the more distributed (non-hierarchical) the network is, the more ways there are to route around failures.

true. but that assumes all nodes would have equal capability to connect to adjacent nodes. i think that's not a good assumption to make, because geographically you may have quite disparate nodes: consider the subway example. you're likely to have many ads clumped together at train stops, but none between stops. you can build long-range networking into all ads, or you could go a cheaper route and hook up the ads to some kind of master node that represents the train stop -- but, in doing so, you've also made the entire stop's ad display dependent on a single node of the network. yet i suppose that, in the end, cost will determine the route that should be taken.

it is impossible to create a network of any real scale with nodes in direct communication with each other because the combinatorics explode too quickly.

i didn't say "direct" communication. i said communication. there is a difference. the ethernet protocol is one in which it is possible to broadcast information to all nodes on the network: in fact, this is necessary to announce the existence of a new node on the network, as the broadcast is a signal to all ethernet nodes to update their routing tables. that is not direct communication, for the broadcast's point of origin itself does not know what computers exist on the network: it relies on consequent broadcasts from other computers to discover who's there.

cryptography and digital signatures should generally be able to solve this problem [of spoofing network nodes].

it will not deter those seriously interested in hacking the network. my biggest problem with wireless networking is that it removes the largest obstacle to joining a network: physically tapping into it. this article is an example of why wireless networking makes me nervous.
posted by moz at 9:17 AM on December 5, 2002

the low-level security of the network doesn't matter if the signs only accept encrypted/signed content. if you don't have the private key, you can't forge a content signature.
posted by muppetboy at 9:46 AM on December 5, 2002

in the end why do you think this stuff is going to be geographically clumped (unless we're talking whole cities)?
posted by muppetboy at 9:51 AM on December 5, 2002

i'm not so convinced re: keys, because can't you crack key encryption? RSA was cracked, and the AirSnort program represents a crack of the 802.11b WEP encryption.

in the end why do you think this stuff is going to be geographically clumped (unless we're talking whole cities)?

it depends on what application we're talking about. subway ads, like i said, will be clumped because the only lighted areas that you aren't zooming by on the train are at train stations. you're going to have a bunch of ads in one place (the stop), and then you may go 5 or 6 blocks before you see anymore ads. every now and then, on an elevated train system, you will see ads along the way; but i've never seen ads along the way that were attached along the line itself, rather signs put on buildings and the like. furthermore, many billboards are located along highways. that would also represent a kind of geographic clumping.
posted by moz at 10:29 AM on December 5, 2002