In his new book, Range: Why Generalists Triumph in a Specialized World,David J. Epstein investigates the significant advantages of generalized cognitive skills for success in a complex world. We’ve heard and read many praises for narrow expertise in both humans and AIs (Watson, Alpha Go, etc.). In both humans and AIs, however, narrow+deep expertise does not translate to adaptiveness when reality presents novel challenges, as it does constantly.
As you ingest this highly readable, non-technical book, please add your observations to the comments below.
Psychologist Robert Epstein, the former editor of Psychology Today, challenges anyone to show the brain processing information or data. The IP metaphor, he says, is so deeply embedded in thinking about thinking it prevents us from learning how the brain really works. Epstein also takes on popular luminaries including Ray Kurzweil and Henry Markram, seeing both exemplifying the extremes of wrongness we get into with the IP metaphor and the notion mental experience could persist outside the organic body.
We’ve had a few discussions about what ‘memes’ are and how they work. We didn’t reach a consensus on whether they are true replicators in their own rights, as genes (or gene collectives) seem to be. Consider the following excerpt (author permission obtained) with the less rigorous working definition of a meme in mind: a unit of information with the capacity to shape perception and belief (or a unit of ‘culture transmission’). Also, consider the conversations and readings we’ve had about metaphors and embodied cognition (particularly Lakoff and Johnson’s works and Edward’s many posts on the topic. Even in Facebook, there are people calling out manipulation by meme.
The NYU Center for Mind, Brain & Consciousness hosts presentations, including topical debates among leading neuroscience researchers. Many of the sessions are recorded for later viewing. The upcoming debate among Joseph LeDoux (Center for Neural Science, NYU), Yaïr Pinto (Psychology, University of Amsterdam), and Elizabeth Schechter (Philosophy, Washington University in St. Louis), will tackle the question, “Do Split-brain patients have two minds?” Previous topics addressed animal consciousness, hierarchical predictive coding and perception, AI ‘machinery,’ AI ethics, unconscious perception, research replication issues, neuroscience and art, explanatory power of mirror neurons, child vs adult learning, and brain-mapping initiatives.
This article was originally published at Aeon and has been republished under Creative Commons.
Cassandra woke up to the rays of the sun streaming through the slats on her blinds, cascading over her naked chest. She stretched, her breasts lifting with her arms as she greeted the sun. She rolled out of bed and put on a shirt, her nipples prominently showing through the thin fabric. She breasted boobily to the stairs, and titted downwards.
This particular hyperbolic gem has been doing the rounds on Tumblr for a while. It resurfaced in April 2018, in response to a viral Twitter challenge posed by the US podcaster Whitney Reynolds: women, describe yourself the way a male writer would.
The dare hit a sweet spot. Many could summon up passages from books containing terrible, sexualised descriptions of women. Some of us recalled Haruki Murakami, whose every novel can be summarised as: ‘Protagonist is an ordinary man, except lots of really beautiful women want to sleep with him.’ Others remembered J M Coetzee, and his variations on the plot: ‘Tenured male professor in English literature sleeps with beautiful female undergraduate.’ It was a way for us to joke about the fact that so much great literature was written by men who could express perfectly detailed visual descriptions of the female body, and yet possessed such an impoverished understanding of the female mind.
This is why the philosophical project of trying to map the contours of other minds needs a reality check. If other humans are beyond our comprehension, what hope is there for understanding the experience of animals, artificial intelligence or aliens?
I am a literature scholar. Over thousands of years of literary history, authors have tried and failed to convey an understanding of Others (with a capital ‘O’). Writing fiction is an exercise that stretches an author’s imagination to its limits. And fiction shows us, again and again, that our capacity to imagine other minds is extremely limited.
It took feminism and postcolonialism to point out that writers were systematically misrepresenting characters who weren’t like them. Male authors, it seems, still struggle to present convincing female characters a lot of the time. The same problem surfaces again when writers try to introduce a figure with a different ethnicity to their own, and fail spectacularly.
I mean, ‘coffee-coloured skin’? Do I really need to find out how much milk you take in the morning to know the ethnicity you have in mind? Writers who keep banging on with food metaphors to describe darker pigmentation show that they don’t appreciate what it’s like to inhabit such skin, nor to have such metaphors applied to it.
Conversely, we recently learnt that some publishers rejected the Korean-American author Leonard Chang’s novel The Lockpicker (2017) – for failing to cater to white readers’ lack of understanding of Korean-Americans. Chang gave ‘none of the details that separate Koreans and Korean-Americans from the rest of us’, one publisher’s letter said. ‘For example, in the scene when she looks into the mirror, you don’t show how she sees her slanted eyes …’ Any failure to understand a nonwhite character, it seems, was the fault of the nonwhite author.
Fiction shows us that nonhuman minds are equally beyond our grasp. Science fiction provides a massive range of the most fanciful depictions of interstellar space travel and communication – but anthropomorphism is rife. Extraterrestrial intelligent life is imagined as Little Green Men (or Little Yellow or Red Men when the author wants to make a particularly crude point about 20th-century geopolitics). Thus alien minds have been subject to the same projections and assumptions that authors have applied to human characters, when they fundamentally differ from the authors themselves.
For instance, let’s look at a meeting of human minds and alien minds. The Chinese science fiction author Liu Cixin is best known for his trilogy starting with The Three-Body Problem (2008). It appeared in English in 2014 and, in that edition, each book has footnotes – because there are some concepts that are simply not translatable from Chinese into English, and English readers need these footnotes to understand what motivates the characters. But there are also aliens in this trilogy. From a different solar system. Yet their motivations don’t need footnoting in translation.
Splendid as the trilogy is, I find that very curious. There is a linguistic-cultural barrier that prevents an understanding of the novel itself, on this planet. Imagine how many footnotes we’d need to really grapple with the motivations of extraterrestrial minds.
Our imaginings of artificial intelligence are similarly dominated by anthropomorphic fantasies. The most common depiction of AI conflates it with robots. AIs are metal men. And it doesn’t matter whether the press is reporting on swarm robots invented in Bristol or a report produced by the House of Lords: the press shall plaster their coverage with Terminator imagery. Unless the men imagining these intelligent robots want to have sex with them, in which case they’re metal women with boobily breasting metal cleavage – a trend spanning the filmic arts from Fritz Lang’s Metropolis (1927) to the contemporary TV series Westworld (2016-). The way that we imagine nonhumans in fiction reflects how little we, as humans, really get each other.
All this supports the idea that embodiment is central to the way we understand one another. The ridiculous situations in which authors miss the mark stem from the difference between the author’s own body and that of the character. It’s hard to imagine what it’s like to be someone else if we can’t feel it. So, much as I enjoyed seeing a woman in high heels outrun a T-Rex in Jurassic World (2015), I knew that the person who came up with that scene clearly has no conception of what it’s like to inhabit a female body, be it human or Tyrannosaurus.
Because stories can teach compassion and empathy, some people argue that we should let AIs read fiction in order to help them understand humans. But I disagree with the idea that compassion and empathy are based on a deep insight into other minds. Sure, some fiction attempts to get us to understand one another. But we don’t need any more than a glimpse of what it’s like to be someone else in order to empathise with them – and, hopefully, to not want to kill and destroy them.
As the US philosopher Thomas Nagel claimed in 1974, a human can’t know what it is like to be a bat, because they are fundamentally alien creatures: their sensory apparatus and their movements are utterly different from ours. But we can imagine ‘segments’, as Nagel wrote. This means that, despite our lack of understanding of bat minds, we can find ways to keep a bat from harm, or even nurse and raise an orphaned baby bat, as cute videos on the internet will show you.
The problem is that sometimes we don’t realise this segment of just a glimpse of something bigger. We don’t realise until a woman, a person of colour, or a dinosaur finds a way to point out the limits of our imagination, and the limits of our understanding. As long as other human minds are beyond our understanding, nonhuman ones certainly are, too.
Kanta Dihal is a postdoctoral research assistant and the research project coordinator of the Leverhulme Centre for the Future of Intelligence at the University of Cambridge.
This article was originally published at Aeon and has been republished under Creative Commons.
Our discussions all, to some extent, relate to cognition. An important area of inquiry concerns whether some form of physical embodiment is required for a brain to support cognition in general and the self-aware sort of cognition we humans possess.
Philosophy In The Flesh: The Embodied Mind And Its Challenge To Western Thought, by George Lakoff and Mark Johnson. Please note, while the title includes “Philosophy,” we are not a philosophy group and the book and discussion will revolve around scientific concepts and implications, not spiritualistic or metaphysical ideas.
– Amazon (used copies in the $6 range, including shipping)
RSVP by email to firstname.lastname@example.org if you plan to attend our discussion on the afternoon of Saturday, November 3, 2018.
While our group enjoys socializing and will plan other events to that end, this meeting is for focused discussion among people who invest the time in advance to inform themselves on the topic. As a courtesy to those who will do their ‘homework,’ before the meeting please read and consider Part 1 (the first eight chapters) of the book. As you read, jot down your thoughts and questions on the book’s claims, supporting evidence, and implications for our core topics–brain, mind, and artificial intelligence. If you are not able to invest this effort prior to the meeting, please do not attend. Thank you for your understanding.
Taken together, these morphological characteristics suggest that neurons in the elephant cortex may synthesize a wider variety of input than the cortical neurons in other mammals.
In terms of cognition, my colleagues and I believe that the integrative cortical circuitry in the elephant supports the idea that they are essentially contemplative animals. Primate brains, by comparison, seem specialized for rapid decision-making and quick reactions to environmental stimuli.
Will new readers develop the more time-demanding cognitive processes nurtured by print-based mediums as they absorb and acquire new cognitive capacities emphasized by digital media? For example, will the combination of reading on digital formats and daily immersion in a variety of digital experiences — from social media to virtual games — impede the formation of the slower cognitive processes, such as critical thinking, personal reflection, imagination, and empathy, that are all part of deep reading?
Wolf first addressed the evolution of reading and its implications in her earlier book, Proust and the Squid: The Story and Science of the Reading Brain. She summarizes her thesis in this interview (14 min video).
Stanislas Dehaene, an author whose work we’ve discussed, also investigated the brain circuits involved in reading. Hear him speak on the topic in this video (33 min).
During our next discussion meeting, we’ll explore the status, future potential, and human implications of neuroprostheses–particularly brain-computer interfaces. If you are local to Albuquerque, check our Meetup announcement to join or RSVP. The announcement text follows.
What are neuroprostheses? How are they used now and what may the future hold for technology-enhanced sensation, motor control, communications, cognition, and other human processes?
Extra Challenge: As you review the resources, think of possible implications from the perspectives of the other topics we’ve recently discussed:
• the dilemma of so much of human opinion and action deriving from non-conscious sources
• questions surrounding what it means to ‘be human’ and what values we place on our notions of humanness (e.g., individuality and social participation, privacy, ‘self-determination’ (or the illusion thereof), organic versus technologically enhanced cognition, etc.)
I like using concept maps to organize my thoughts and to plan articles and projects. I would like to share member-editable concept maps through this website.
I’ve used the free Cmap Tools program for many years. Unfortunately, Cmap Tools is developed on the Java platform and I have been unable to get it to work on my Mac for about two years. (Apparently, it needs an older version of the Java Runtime Environment and several other tools I use need a later version, which creates a conflict I’ve not been able to resolve.)
I am able to use Cmap Tools on my Windows 10 virtual machine, so it’s still a good candidate application.
Fortunately, there’s also a free cloud version of Cmap we can use to share concept maps. The downloadable (client) application has many more features for embedding functionality into concept maps, but the cloud environment is excellent for viewing and basic editing. Here’s a sample cmap I built in about five minutes this morning (displayed using the embed code from the Cmap Cloud Viewer):
Because this example is presented in a viewer window, it is not directly editable. If I provide a link to a shared cmap online, you’ll be able to edit it as well. If you install the downloadable version of Cmap Tools, you’ll be able to add richer content and more advanced features, including collapsable detailed nodes, embedded images and other resources, links to other nodes and maps, links to online content, etc.
If you might be interested in viewing and/or collaborating on concept maps, please reply and identify the operating system of the computer you would use. (If you don’t want to post your OS reply, please tell me by email.) Once you create a free account online, we’ll be able to collaboratively work on cmaps, which is a great way to negotiate meaning with a bit more structure and persistence than real-time conversations.
Last, if you want to know how cmaps and other visual representation benefit learning, read the articles on the IHMC site. I also highly recommend the little book, Learning How to Learn, by Novak and Gowin. Thanks!