Wednesday, November 23, 2022

The Frog's Eye and the Feedback Loop



On a sunny summer day, a Northern Leopard frog sits on a lily pad at the edge of a pond.  A beetle lands nearby and retracts its wings.  After a moment of stillness, it starts to crawl.  The frog, immobile to now, leaps on the beetle in a flash, scoops it into its mouth with its sticky tongue and swallows.  It returns to stillness.  Moments later, the shadow of a heron wading close by falls suddenly over the frog.  Before the heron has a chance to swoop in, the frog leaps toward a patch of darkness on the surface of the pond and escapes under the canopy of bobbing lily pads swimming for deeper water.  It is a simple sequence of interactions, facilitated by the frog's vision.

The eyes of a frog protrude from its head to a unique extent for any vertebrate.  They are relatively immobile, moving only as the frog moves its head.  Unlike human eyes which have a focal point that requires them to be directed at an object in order to see it, every surface of the frog's eye sees equally all at once, in light and, masterfully, in dark.  Even a frog's eyelids-- a top, a bottom and a third nictating eyelid that lubricates the eye when it is on land-- are transparent.  Moreover due to parallax effects of the placement of the eyes and their bulging shape a frog has nearly 360° of peripheral vision at all times.  Thus a frog is a vision machine, turned on whenever the frog is awake, and evidence suggests even when it's not. Frogs see color and as has been recently demonstrated can even distinguish between shades in what a human would consider complete darkness.  In fact, a frog's survival is very much dependent on what its eyes can perceive.  As one researcher put it, a frog "will starve to death surrounded by food if it is not moving." 

Our intuition about vision is that the eyes merely perceive the external world by transmitting images whole to the brain to interpret. But this is not the case for frogs.  

We know this because of a landmark study of frog vision by Jerome Lettvin, Humberto Maturana, W.S. McCulloch and Will Pitts, described in their paper, "What the frog's eye tells the frog's brain"  published in 1959 for the Proceedings of the Institute of Radio Engineers.  Lettvin a professor of electrical and bioengineering and communications physiology at MIT had met Pitts-- a self-taught wunderkind in math and logic from Detroit--  as a graduate student at University of Illinois. Lettvin, Pitts and their American colleague McCulloch were all researchers in the up and coming field of cybernetics, the study of the organization and control of systems biological and electronic, and especially the interaction between machines-- specifically computers-- and the humans who operate and interact with them. The term 'Cybernetics' was coined in 1947 from the Greek κυβερνήτης for "steersman" by Norbert Wiener and Arturo Rosenblueth, early pioneers in the field.  Wiener and his group at MIT were interested in studying components of systems (mechanical, organizational and biological) built to respond to problems organically encountered in the course of operation-- a class of engineering problem the solution to which promised a wide array of applications, not least the engineering  of what has come to be thought of as artificial intelligence.  The bedrock of cybernetics was the feedback loop, in which one part of a system provides input to a processor producing a response which in turn is intepreted by the originating component as instructions for what to provide next in the way of input.  

Ensconced at MIT following his graduate work, Lettvin had become interested in vision and the role it played in how creatures interact with the world. Maturana, a Chilean national completing his Ph.D. at Harvard at the same time had done electron microscopy of the hundreds of thousands of fibers in the frog's optic nerve and his results had inspired the method devised by the researchers for distinguishing classes of nerve bundles in order to intercept neural firings to get readings of responses to visual stimuli.

For the experiment which used the Northern Leopard frog (Rana pipiens) as subjects, electrodes were attached strategically to the optic nerve and to the superior colliculus-- the vision receptor of the brain.  Over the eye was mounted a specially constructed aluminum hemisphere on the surface of which were moved various objects of different shapes, sizes and colors by means of magnets in varying light.  In this way, readings could be taken of the frog's neural responses to the visual stimuli.  What the researchers discovered was that certain bundles of nerve fibers were prone to respond to specific categories of visual stimulus.  Contrast between contiguous areas of the visual field (distinction of an object from the area surrounding it) was detected by one type of bundle.  The movement of an edge on the plane evoked responses from another. The net curvature of an edge (the shape of much of a frog's diet) stimulated a third category of bundle. Another was sensitive to changes in degree of ambient darkness-- such as what might be caused by a descending shadow.  Each type of bundle transmitted a signature pattern of impulses to the brain at the introduction of its associated visual stimulus.  The researchers concluded that the optic system of the frog had evolved to detect specific features of the visual horizon in whole, and that far from being a passive camera scanning the environment for images for the brain to interpret was on its own detecting and sending information whole pre-interpreted for the brain for a hastened response-- capture and intake in the case of beetles, flight in the case of herons.  As Lettvin put it to a colleague, “The eye speaks to the brain in a language already highly organized and interpreted, instead of transmitting some more or less accurate copy of the distribution of light on the receptors.”

 We're told that Lettvin was laughed off the stage at American Physiology Society in Atlantic City when he presented the paper.  Walter Rosenblith, an MIT colleague and leading researcher in vision was so skeptical of Lettvin's results that he snubbed his colleague in assembling a meeting on vision shortly after the conference-- but on challenging Lettvin in his lab to demonstrate his methods, he was reportedly convinced.  The paper became among the most cited for at least the next ten years. 

One researcher inspired by it was Stafford Beer.   Beer was a British theorist, consultant and author in business management of a field that he called Operational Research.  Heavily influenced by the cybernetics of Norbert Wiener he had written for years about how the optimal structure of an organization should emulate the nervous system of an organism -- with tendrils into the environment-to-be-perceived and fed to a central brain for processing the data and quickly directing the limbs in how to respond to conditions at hand, but his focus had been on the structure of the midcentury corporation.  He had never had an opportunity to apply his theories on anything beyond a client company, let alone to a social movement, until he was approached by Fernando Flores, a young up-and-comer in Salvador Allende's coalition Socialist government that had been elected in Chile in November 1970.  

As a graduate student at the University of Illinois in the early 60s, Flores had been inspired by encounters with the cybernetics community headed  by Heinz von Foerster, director of the Biological Computer Laboratory.  Now working for Allende's government as an employee of  CORFO, the acronym for Chilean Production Development Corporation, Flores wanted a system cybernetically devised to permit rapid response to current conditions within the various economic sectors of the country.  To this end he hired Beer as consultant to engineer the project.  

Beer had been thinking for years about how to get around the persistent problem in pre-internet times of information lag.  Owing to the speed at which economic data was collected, typed up and published, information  gathered for strategic planning purposes was typically 3 months old before it was analyzed by decision makers, making the response frequently too late and irrelevant to the situation on the ground.  Given the international opposition to Allende's election immediately raised from the world's most powerful players economically and strategically,  opposition in the form of barriers of trade and blacklisting (to say nothing of the active plotting against Allende orchestrated and enabled from the Nixon administration and CIA, including the assassination of a pro-Allende General shortly after the 1970 election in an unsuccessful attempt to embolden anti-Allende factions in the military to prevent Allende's assumption of the office), Flores was wanting to devise for Chile something that did not exist at the time anywhere in the world except in the US and UK, and at that, still very much as more a set of ad hoc links set up between disparate technologies of a handful of universities than an integrated system.  

In a very real sense, Beer was tasked with creating an interconnected network-- an internet if you will-- from the ground up.  The project was dubbed Cybersyn (a portmanteau of Cybernetics and Synergy) in English and Synco (for Sistema de Información y Control) in Spanish.  Beer's original design called for links between the behemoth mainframes of the time strategically placed around Chile's challenging geography (4000 miles from North to South and only 60 miles from East to West.), but at the time only one was available for the project's use in the country, an IBM/3060 in Santiago-- and given the hostility of the international business community to Allende's socialist  government, likely to be the only for the foreseeable future. 

As Beer discovered, one thing that Chile had no shortage of at the time was Telexes, hundreds of which had been lying in storage since the project for which they were originally procured had been shelved by a previous administration.  The windfall permitted instant communication of data over great distances through telephone lines. Telexes were installed in strategic factories, farms and offices across the country.  To gather and process the information, Beer conceived of a "war room" of large-screen wall-mounted monitors fed with data and controlled by decision makers arranged in a circle to enhance the democratic spirit of collaboration in specially designed fiberglass chairs custom made in Chile per Beer's specifications. with simplified controls built into the arms giving each participant the ability to call up charts and relevant data from their respective sectors for group discussion.  The technology was new but the war room bore a certain resemblance to something out of Star Trek or Kubrick's 2001.  

With a middling economy that had historically underperformed, Chile was an unlikely place for an internet to be pioneered.  In devising a system to maintain Cybersyn beyond his involvement with the project, Beer took inspiration from the new theory of 'autopoiesis' describing the self-regeneration of biological -- and by extension, human designed-- systems that was developed in Chile by none other than Humberto Maturana with his colleague at University of Chile Francisco Varela.  Beer invited Maturana and the latter's University of Illinois mentor Heinz von Foerster to see the progress of the project and share thoughts and suggestions with the team.  Allende himself visited.  

The strikingly futuristic room was small and the heat generated from the cramped equipment and close quarters sometimes caused malfunctions, but the Cybersyn concept was given a test by a paralyzing national trucker strike joined by small business owners and shopkeepers in September 1972 (all of it instigated and funded covertly by Nixon and Kissinger's CIA as it turned out) to bring their foe Allende's economy to a halt.  As a test of Cybersyn's design and purpose, word from across the country was quickly brought to the war room via Telex. In this manner the team could rapidly determine where there were shortages, where there might be factories that Allende's government could nationalize.  In spite of the striking of the petit bourgeoisie, workers who were very much in the Allende camp assumed control of factories and with the help of the cybersyn team knew where to send goods and equipment to keep the country fed and supplied with essentials.  The strike which ended after 26 days was weathered successfully in part due to the part that Cybersyn played in mitigating stoppages and blockades.

A second strike of shopkeepers, professionals and Santiago taxi drivers in June 1973 (also enabled due to heavy CIA involvement) further destabilized the country.  On September 11, a violent military coup headed by General Augusto Pinochet seized control of the government.  In the course of events, Allende was killed at the Presidential Palace (by suicide a scientific autopsy much later appeared to confirm).  Stafford Beer was back in the UK at the time.  Fernando Flores,  who had met with Salvador Allende earlier that day was later arrested along with hundreds of Allende officials, imprisoned and subjected to the Pinochet regime's brutal brand of torture for 2 years before being released and forced into exile.* 

In spite of its early successes, Project Cybersyn's fate had already been somewhat sealed. In the face of growing economic and societal instability, intense right wing propaganda (and to some extent poor PR on the part of the Allende government), the project had come to be tarred as authoritarian overreach.  The taint of its association with Allende's moderate socialism all but guaranteed that it would be relegated to the dustbin of history by Pinochet's extreme right-wing coup.  Heavily backed by the US,  Chile became a laboratory for the austere economic policies of the University of Chicago school personified by Milton Friedman, an early testing ground for the neoliberal world order that would soon come to dominate and suffocate the global economic and political order. 

The public's introduction to Cybersyn was somewhat responsible for its demise.  To the Cybersyn team, Control didn't mean Authoritarian rule as was frequently misinterpreted to great effect by foes of Allende's project, but rather informed response to the best information.   In fact, when Stafford Beer was tasked to sell Salvador Allende on the plan, in explaining one of his famous charts of his Viable System Model of overlapping circles demonstrating the flow of information and control in a system in all directions, he was struck by Allende's characterization of the highest level of the hierarchy, the brain of the organism as being "por fin, el pueblo" (At last the people).  In the conception of the Cybersyn team, the fuel for the feedback loop was information transmitted, fully interpreted, by the specialized organ of the state's economy, the workers, to stimulate an informed response from a centralized decision-making organ acting to the benefit of the body politic.  Much like the eyes and brain of the frog.  


Georgios Cherouvim

_____
Inspired by material from Cybernetic Revolutionaries, by Eden Medina, MIT Press, 2011.

~~~~~~
* On his release, with the Pinochet government still in full swing, Flores, whose brainchild Project Cybersyn was, emigrated to the US where he took up residence at Stanford University. Finding himself present at the creation of Silicon Valley, he was able to ride the wave of the earliest internet bubbles to turn his experience and expertise in organizational logistics into a fortune-- ironically becoming an exponent of Silicon Valley libertarianism.  For Stafford Beer, Cybersyn was a point on the path of an opposite trajectory-- the one-time business consultant was turned into a committed socialist by his experience of Allende's Chile; and he continued to offer his services to cybernetic projects in Mexico, Venezuela and Uruguay.

No comments:

Post a Comment