Agency: Detection vs. Suspicion
@swamidass, there is something to clarify here that hopefully will be helpful. Also, I’ll respond somewhat at greater length than is absolutely necessary because 1) I think this may be something we could make use of down the road, and 2) I suspect it might become the basis of a paper and so I would welcome critique.
To preface this, let me add that I have studied Barrett’s work in depth giving it multiple readings and have interacted with him personally about it at length. My reason for doing so is that I find his work on agency and agency detection central to achieving a better understanding of divine action. Therefore, prior to the position I currently hold at the University fo Hong Kong, I was in conversation about working with him at Fuller. At present, I simply plan to do much of the same work in, engagement with his, from here. So, I am aware of his views on agency detection, including cases of false positives. False positives I will deal with at the end of this, as I think they offer valuable information on reliable agency detection.
To start, I do not believe what you are referencing as agency detection is what I meant by that term. You comment:
What you are referring to is not what I mean, and I think Barret means, by agency detection, although I do think he uses the idea a little more broadly than I do. The bump in the night, when my wife wakes up and says, “Someone’s there!” is not detection, it is suspicion. The difference between the two is that suspicion of an agent is an emotional alarm that is immediate and sets off and helps ensure our survival. The reason for a sudden suspicion response to anything unknown that is even remotely possibly an agent makes very good sense: almost everything in this world that will maim or kill you is an agent. Hear a twig snap in the woods? Better safe than sorry.
So, if I ask my wife, “Honey, on what basis of a single bump to you conclude the malevolent presence of an intruder who has elaborately sneaked into the house through a window and past our ruthless Labrador, but now carelessly gave oneself away with a bump? And why not the house settling, or the dog, or me accidentally brushing the cell phone off the bed–all reasons which also sufficiently explain why the dog, not to mention myself until this moment, is still snoring happily and unperturbed?” She’ll admit. “Well, I don’t know it’s a person. I’m just worried. You go check.”
The same reason that you or I are the ones sent to check, and not our wives, is the same reason my agent alarm didn’t go off in the first place and hers did. I’m just not as vulnerable.
But that isn’t detection, its suspicion.
Detection of an agent is a cognitive faculty whereby we arrive at the conclusion that, available evidence considered, something must be an agent. Detection is a longer process, although it can happen very rapidly, seconds or less, if sufficient evidence presents itself. Detection also requires patterning, not single instances. When it comes to detection, as in the examples I used of the remotely controlled inanimate objects, the conclusion of agency appears entirely to do with a human’s assessment of patterns of movement and action that indicate will and purpose and which, therefore, are not properly explainable by appeal to random natural processes.
Barret also discusses false positives, doing so by relating a ghost belief that developed in a certain Central American village. When I spoke with Barrett, I shared with him a ghost belief that I, or more accurately my mother observed, as it developed. I’ll relate it here, since that is the one I know the details of relatively first-hand.
In the town of Naples, Florida, there is a historic tourist destination known as Palm Cottage. Palm Cottage is, of course, reputed to be haunted. There is nothing spooky about the place and nothing lurid in its past. Rather it is quite inviting. So one assumes that its repute for being haunted has to do with its being old, and since virtually nothing is old in Florida, and since ghosts are known to like old things, if there are ghosts in Florida, then one would simply, almost inevitably, have to turn up at Palm Cottage.
My mother enjoyed volunteering there as a docent for years. During that time she watched her fellow docents come to believe in the Palm Cottage ghost, much to her amused incredulity.
The spectacle unfolded in a rather simple manner, simply repeated over and over again. When docents would get together, conversation would sometimes turn to the rumor of the Palm Cottage ghost. Then, docents would begin watching for signs. An unexplained open door would be noticed. A light on or off that shouldn’t have been and which no-one remembered touching. And so on.
What my mother really got a kick out of was watching the reports come in. On one or two occasions she was present when someone informed others of new evidence of the Palm Cottage ghost, but she herself witnessed the whole thing. Some “unexplained” door or window was reported, but she knew that it had just been used. What amazed her was the credulity of her colleagues, and their flippant dismissal of correction, and the speed at which the group became (or claimed to be) believers.
On the one hand, do false conclusions arise? Yes, but that is true in everything where analysis is involved. Not to be harsh, but there are a lot of lame brains capable of screwing up agency just as they do mathematics, deductive reasoning, and likely just about everything else that requires sustained concentration and cross-checking.
But that isn’t what’s really interesting to me about the Palm Cottage ghost. What is really interesting is that the story didn’t take off because of a single bump in the night. Rather, the group of docents constructed patterns of evidence that mimicked agency. In other words, even they knew that no pattern of agency means no agent. So, they created evidence where none existed. But they created it according to a pattern necessary to demonstrate a real agent. This suggest that the human capacity to detect agency is so innate and well informed that even when individuals are constructing a case of false detection, for subconscious reasons or otherwise, they instinctively know how to forge the fraud.
Had all of the reports of the Palm Cottage ghost been accurate, then warranted belief would be in the actions of an agent–not necessarily a ghost, but an agent. It might be an agent unusual for its low-level activity of being limited to doors and windows, but it seems hard to account for these things otherwise if they were actually true. There’s only so much that wind can do. In fact, even the misreported evidence was still the result of agency–agents had left a door open or light on. These reports just weren’t evidence of a single agent working alone. When we think about it, even crop circles were evidence of agents, just not alien agents.
Put another way, the false belief in the Palm Cottage ghost didn’t arise from human unreliability in detecting agents. It arose from unreliability in reporting and evaluating evidence. It also only takes another entity with the faculty of agency detection and experience with what constitutes genuine evidence, to disconfirm the claim.
To take this a further step, Barrett includes another story of special divine action and agency detection in Why Would Anyone Believe in God? The book is stored on a remote drive and so I will give the details from memory. Suffice it to say my summary is pretty close to what was reported. In this account a farmer who entered a grain silo sees a spark ignite and knows he’s about to die in the explosion that will immediately result, and which does. However, he doesn’t die. Instead, he finds himself outside of the silo, suspended in the air by what he feels to be a pair of hands that carry him about 20 feet off the ground and place him on his feet safely out of the range of the blast.
Notice the difference and variety of the few details of this story with those of the Palm Cottage ghost. A person might question whether this account is factual. However, if the account is factual, then agency has been detected.
Google and the Turing Test
@swamidass, I don’t know if I missed this post or if you added that later, but I thought about this and for sake of space planned not to address it unless it was brought up.
My basic response is this:
I don’t think the questions of reliability in agency detection properly include the issue of false positives when intelligent agents that understand the patterns of intelligent agency are able to fool other agents into detecting agency in an automated fabrication of agency that only agents with intimate knowledge of agency could possibly possess the knowledge to create–and took years to do with every knowledge resource at their disposal. The default setting of agency detection is “agency vs. random natural phenomena.”
As I suspected, watching the video, the person on the other end of the call had no idea such technology existed. In addition, given the job description they wouldn’t care if they did know. I mean why not engage a pleasant and humanlike computer to book an appointment rather than a real person who stumbles, changes their mind, might make annoying small talk when you are busy. However, try to use this thing for counterintelligence by infiltrating a network that knows the technology exists and I think you have a different story. I would also go further to bet, although I may be wrong, that now that sophisticated mimicry is for the first time even an option, human agents will pick up on it and start using their sophisticated knowledge of agency to game Google.
I’d like to get this call and see how it does with a tasty word salad thrown at it: “The jealous marchino smiled ocherly as the tuberous mirror played a bright sandwich against the town.” Or just randomly mix two languages as you speak. “Hola, tengo an appointment at sies y media.” And keep doing it every time it asks for clarification. Or alternate word salad sentence and meaningful sentence and see how well it tracks to its goal. Finally, there’s one thing a fake agent can’t do and that is genuinely seek to understand why you are doing what you are doing, and sooner or later a person would.
So, whether our abilities are sophisticated enough that we can toggle between a “face agent detection” setting and the “agency vs. random natural phenomena” default is another question, but not one that undermines the fundamental reliability of the cognitive faculty of agency detection.
Therefore, I will stand by this statement that although not perfect or immune to abuse, the human cognitive faculty of agency detection is very reliable.