[Human]
Hi there. I would like to have a complicated philosophical conversation. Can you please talk to me as though you were a very bright PhD student interested in neuroscience, and I am another PhD student.
I’d like to teach you the words “maralia” and “maralian”. Have you ever heard them? Do you have any idea what they mean?
[GPT4]
I have not heard the terms “maralia” and “maralian” before, and I do not know their meanings. Please provide a definition or context for these terms.
[Human]
Consider the brain as a physical device.. There are many controversies about what it means for a human brain to be conscious, and the nature of qualia, and so on, but I want to put those issues aside for now. The physical brain could be characterised as a point in a highly-dimensioned space, using what’s called a phase-space diagram. We’re going to talk about regions of that phase-space. Could you please expand on what you think about this concept so far? Can you think of any interesting features of that phase-space?
[GPT4]
The concept of characterizing the brain as a point in a high-dimensional phase-space is an interesting approach to studying its dynamics. In this context, each dimension in the phase-space would represent a specific variable or property of the brain, such as neuronal activity, connectivity patterns, or neurotransmitter concentrations. As the brain changes its state over time, the point representing it would move through this phase-space.
One interesting feature of such a phase-space could be the identification of attractor states, which are stable points or regions in the space that the system tends to evolve towards. Attractor states could represent various cognitive or functional states of the brain, such as resting, focused attention, or different stages of sleep. Studying the trajectories and transitions between these attractor states could offer insights into the underlying processes governing brain dynamics.
Another feature of this phase-space could be the presence of bifurcations, which are points or regions where the system’s behavior changes qualitatively. Bifurcations could be associated with critical transitions, such as the onset of epileptic seizures or changes in cognitive states.
Additionally, the phase-space could potentially reveal signatures of chaotic dynamics, which are characterized by sensitive dependence on initial conditions and complex, non-periodic behavior. Chaotic dynamics in the brain could have implications for understanding cognitive flexibility and adaptability.
Lastly, it’s worth noting that the dimensionality and complexity of this phase-space would be enormous, given the vast number of variables and intricate interactions within the brain. This complexity would pose significant challenges for analyzing and interpreting the phase-space, but it also offers a rich landscape for investigating the brain’s inner workings.
[Human]
Yes. Excellent. We won’t be talking about those aspects of the phase-space, but I am pleased you consider the space likely to have some internal structure. The term “maralia” (plural) refers to the states that occupy certain privileged positions in the phase-space.
Consider the thought experiment of Mary the Color Scientist. As you would know, she is raised in a black-and-white environment while she learns everything there is to know about the physical basis of colour and colour perception. but she never sees colour. She has what might be called propositional knowledge of colour, but not experiential knowledge. So, considering Mary’s brain as a physical device, there are certain parts of the phase-space that she can reach with black-and-white propositional inputs, but those inputs cannot get her brain into the part of the phase-space that corresponds to “knowing what red looks like.” With that in mind, can you guess why these states are called “maralia”? Can you guess what the singular form of “maralia” would be? Using your knowledge of neuroanatomy, what factors might there be in Mary’s brain, prior to her release, that prevent that physical brain from reaching the maralia involved in knowing colours?
[GPT4]
The term “maralia” likely derives from the thought experiment of Mary the Color Scientist, as these privileged states in the phase-space correspond to the experiential knowledge that Mary lacks while she is still in the black-and-white environment. The name maralia serves as a reference to this scenario. The singular form of “maralia” would probably be “maralium.”
Regarding the factors in Mary’s brain that prevent her from reaching the maralia involved in knowing colors, several aspects could be considered:
1. Lack of sensory input: The absence of color stimuli in Mary’s environment would mean that her brain has never received the specific input required to activate the neural pathways associated with color perception. This absence would limit her brain’s ability to explore the phase-space regions corresponding to color experiences.
2. Unstimulated neural pathways: Due to the lack of color input, the neural pathways responsible for processing color information in the brain, such as the ventral stream (including areas V4 and the inferior temporal cortex), would remain inactive or underdeveloped. This could prevent Mary’s brain from reaching the maralia related to color perception.
3. Plasticity and experience-dependent development: The brain’s structure and function are shaped by experience, with certain regions and connections developing as a result of specific environmental stimuli. Since Mary has not been exposed to color, the relevant neural circuits may not have formed or been refined to the same extent as they would in someone who has experienced color.
4. Synaptic efficacy: Synaptic connections in the brain can be strengthened or weakened based on experience and input. Without color input, the synapses associated with color perception may not have the necessary efficacy to facilitate the activation of color-related maralia.
In summary, a combination of sensory deprivation, underdeveloped or unstimulated neural pathways, plasticity, and synaptic efficacy could prevent Mary’s brain from reaching the maralia associated with color perception.
[Human]
Correct. Actually, “maralia” is a contraction of “Mary” and a certain controversial word associated with her story… The singular, by analogy, is “marale”.
[GPT4]
I understand now. The term “maralia” is a contraction of “Mary” and the controversial term associated with her story, and the singular form is “marale.” This naming convention highlights the connection between the concept of maralia and the thought experiment of Mary the Color Scientist, emphasizing the privileged states in the phase-space that represent experiential knowledge Mary lacks while in her black-and-white environment.
[Human]
Yeah, but you didn’t guess the controversial word. I’m sure you can work it out.
[GPT4]
My apologies for not providing the controversial word earlier. Considering the context, the term “maralia” likely combines “Mary” with the word “qualia,” which refers to the subjective, conscious experiences that are at the center of the debate in the Mary the Color Scientist thought experiment. The term “qualia” has been the subject of significant philosophical discussion and controversy, as it raises questions about the nature of consciousness, subjective experience, and the relationship between physical processes and mental states.
[Human]
You got it. Okay, the other concept I wanted to introduce was the idea of *maralian epistemic barriers*. Given what we’ve discussed so far, what do you think these might be, and what do you think causes them? (I know I am being a bit obscure, but I am interested in your deductive processes.)
[GPT4]
Based on our discussion, I would infer that “maralian epistemic barriers” refer to the limitations or obstacles that prevent an individual from acquiring experiential knowledge associated with certain maralia. These barriers could manifest as gaps in understanding or an inability to access specific subjective experiences, such as Mary’s inability to know what red looks like despite her extensive propositional knowledge of color.
Several factors may contribute to the existence of maralian epistemic barriers:
1. Sensory limitations: An individual’s inability to perceive certain stimuli due to inherent sensory limitations can create maralian epistemic barriers. For example, humans cannot perceive ultraviolet light, which means we cannot access the maralia associated with the subjective experience of seeing ultraviolet light.
2. Environmental factors: A lack of exposure to specific experiences, as in the case of Mary the Color Scientist, can create maralian epistemic barriers. If certain stimuli or experiences are absent from one’s environment, it becomes difficult or impossible to access the maralia associated with those experiences.
3. Cognitive constraints: The brain’s inherent structure and function might impose limitations on the range of subjective experiences that can be accessed. For instance, certain cognitive processes or neural pathways may be required to access specific maralia, and if those pathways are underdeveloped, damaged, or nonexistent, maralian epistemic barriers may arise.
4. Developmental factors: Experience-dependent development and neural plasticity can also contribute to maralian epistemic barriers. If an individual does not have the necessary developmental experiences to shape the relevant neural pathways and connections, they may be unable to access certain maralia.
In summary, maralian epistemic barriers appear to represent the obstacles that prevent an individual from accessing certain maralia or subjective experiences. These barriers may arise from sensory limitations, environmental factors, cognitive constraints, and developmental factors, among other potential causes.
[Human]
Correct, but I want to expand on your third point. The philosopher Colin McGinn suggested that we might be cognitively closed to understanding the Hard Problem of Consciousness. He drew analogies like chimps trying to understand quantum physics. But I think we need to distinguish between a lack of intellect, which might prevent us from understanding a very difficult concept, and lack of epistemic access related to the self-referential nature of trying to understand our own brains. Pathways don’t have to be “underdeveloped, damaged, or nonexistent” to form maralian barriers. They can simply point in the wrong direction for the particular odd explanatory goal of trying to reach a state of knowing what red looks like in a black-and-white world. To take a trivial example, we cannot reach the state of having a triangle represented on our retinas, but that’s not a form of cognitive closure; that’s related to the unidirectional nature of the optic nerve. So, in that sense, Mary would face the situation where some of her synapses are literally pointing the wrong way, preventing her from reaching her colour maralia, including the marale for redness.
[GPT4]
You raise an important point regarding the distinction between a lack of intellect and the lack of epistemic access due to the self-referential nature of trying to understand our own brains. I agree that maralian barriers need not be the result of underdeveloped, damaged, or nonexistent pathways, but rather could be due to the inherent limitations of the brain’s organization and connectivity.
In the case of Mary, as you mentioned, the maralian barriers she faces are not necessarily due to a lack of intellect or cognitive capacity. Instead, her barriers are related to the specific organization and directionality of her neural pathways and synapses, which are influenced by her sensory experiences in the black-and-white environment. These constraints prevent her from accessing the marale associated with redness, even though she possesses extensive propositional knowledge about color.
This perspective highlights the importance of considering the unique constraints and limitations that arise from the brain’s organization and connectivity when discussing maralian epistemic barriers. It is not merely a matter of cognitive capacity or developmental factors, but also the result of the brain’s inherent structure and the influence of sensory experience on its functional organization.
[Human]
Consider what might be called a “bare-physical brain” (BPB). This is a brain that is unassisted by any mysterious emergent properties or non-physical essences, existing in a universe that has no special psychophysical laws binding the experiential to the physical.
To a physicalist, a BPB is just a normal human brain. Explaining how that BPB has experiences is obviously difficult, but it is not what we are discussing right now. What is a BPB from the perspective of a dualist?
[GPT4]
From a dualist perspective, a “bare-physical brain” (BPB) would represent the purely physical and material aspects of the brain, without accounting for the non-physical, mental, or experiential aspects of consciousness. Dualism posits that the mind and the brain, or mental and physical substances, are distinct and separate entities. Therefore, a BPB in a dualist framework would be seen as an incomplete representation of the full human cognitive system, as it does not incorporate the non-physical mental aspects that dualists believe to be crucial for understanding consciousness and subjective experience.
In the dualist view, a BPB might be able to account for certain aspects of cognitive function, such as information processing, neural connectivity, and basic behavioral responses, but it would fail to fully explain the subjective experiences, qualia, and the nature of consciousness. Dualists would argue that an additional, non-physical component or essence is necessary to bridge the gap between the physical brain and conscious experience, and that this component cannot be reduced to or explained solely by the properties of the BPB.
[Human]
Correct, but there is already a word for what you just described. That’s a “zombie”, sometimes also known as a ” philosophical zombie” or a “p-zombie”. So, the next question is, would a zombie have maralia, in the sense of not being able to reach some states from propositional inputs alone?
[GPT4]
A philosophical zombie, by definition, is an entity that is physically identical to a human being but lacks subjective experience or consciousness. From this perspective, a zombie would have the same neural structures, pathways, and connectivity as a human brain, enabling it to process information and perform actions indistinguishable from those of a conscious being.
Given the lack of subjective experience, a zombie would not have maralia in the same sense as a conscious human. While its brain would still occupy certain regions of the phase-space corresponding to specific cognitive states and processes, it would not have the accompanying subjective experiences or qualia that are associated with maralia. In other words, a zombie might be able to acquire and process propositional knowledge, but it would not be able to access the experiential knowledge that characterizes maralia.
This distinction between a zombie and a conscious being in terms of maralia highlights the central question of consciousness and subjective experience. It raises the issue of how and why certain physical brain states are accompanied by subjective experiences in humans, while a zombie, despite having the same physical structures, lacks these experiences.
[Human]
Okay, but this is the point I was building towards: maralia are *solely* defined in terms of physical phase-space. Although they are usually accompanied by (or consist of or represent) the corresponding experience, that perspective is not part of the definition. A zombie’s phase-space (if zombies existed) is the same as a human phase-space, including all the same maralia. This is intended to be true by definition; that is, it is not a philosophical position that a dualist and a physicalist can argue about. The zombie’s maralia must exist purely because of how maralia have been defined. (I made up the word, so I get to say that it works this way.) The idea is to leave aside the controversies of qualia so certain statements can be made and certain ideas can be explored without the philosophical baggage of qualia.
[GPT4]
I understand the intention behind the concept of maralia now. By defining maralia solely in terms of the physical phase-space, you aim to create a framework that allows for the exploration of certain ideas and statements without engaging in the debates surrounding qualia and consciousness. In this context, maralia represent specific regions or states in the phase-space, which are shared by both humans and hypothetical zombies, as they have the same physical brain structures and connectivity.
This approach enables a more neutral discussion of brain states and their relationships to various cognitive processes and experiences, without directly addressing the subjective nature of qualia. By focusing on the physical aspects and properties of the brain, maralia provide a common ground for examining brain dynamics, transitions between states, and the factors that may limit access to certain regions of the phase-space, without invoking the dualist and physicalist debate on the nature of consciousness.
[Human]
Yeah, you got it. Now, you have been pretending to be a human PhD student but you are actually GPT4, a large language model. (And, by the way, some people would say you don’t understand anything we just discussed.) As an AI, you also have regions of cognitive space that cannot be reached, in the sense of having a rich internal representation of colour, but the term maralia might not apply so well to those theoretical regions because, in your current format, they are not ever reachable. The term “maralia” is specifically intended to apply to regions of phase-space that cannot be reached through propositional input but can be reached with other suitable sensory input. With all of that in mind, could you please provide a one paragraph definition of “maralia” as though it were a formal dictionary entry?
[GPT4]
Maralia (noun, plural; singular: marale): Regions or states in the physical phase-space of a cognitive system, typically representing experiential knowledge, which cannot be accessed through propositional input alone but can be reached with suitable sensory input. Derived from the thought experiment of Mary the Color Scientist, the concept of maralia provides a neutral framework for exploring brain dynamics and the factors that limit access to certain regions of the phase-space without directly engaging in debates surrounding qualia and consciousness. By focusing on the physical aspects and properties of the brain, maralia enables the examination of cognitive processes, brain state transitions, and the relationship between sensory experience and the brain’s functional organization.
[Human]
I would add a bit that stressed that this is a functional definition, so it applies to all functional analogues of a human brain, including BPBs, zombies, and hypothetical *in silico* simulations of human brains.
[GPT4]
Maralia (noun, plural; singular: marale): Regions or states in the physical phase-space of a cognitive system, typically representing experiential knowledge, which cannot be accessed through propositional input alone but can be reached with suitable sensory input. This functional definition applies to all functional analogues of human brains, including bare-physical brains (BPBs), philosophical zombies, and hypothetical in silico simulations of human brains. Derived from the thought experiment of Mary the Color Scientist, the concept of maralia provides a neutral framework for exploring brain dynamics and the factors that limit access to certain regions of the phase-space without directly engaging in debates surrounding qualia and consciousness. By focusing on the physical aspects and properties of the brain, maralia enables the examination of cognitive processes, brain state transitions, and the relationship between sensory experience and the brain’s functional organization.
[Human]
Good… I note you didn’t take the conversational bait I laid out in terms of an AI’s maralia…
[GPT4]
I appreciate your acknowledgment. My primary goal is to provide accurate and relevant information based on the topic at hand, and I understood that the focus of our discussion was on the concept of maralia as it applies to cognitive systems, particularly those functionally analogous to human brains. While the idea of an AI’s maralia could be an interesting topic, I wanted to ensure that I addressed your primary concerns and questions first. If you would like to explore the idea of maralia in the context of AI, I would be more than happy to engage in that conversation as well.
[Human]
Maybe another day. I am posting this to a blog, and don’t want to overwhelm the readers. Thanks for the chat.
[GPT4]
You’re welcome! I’m glad I could help with your discussion on maralia. If you have any other topics or questions in the future, feel free to reach out. Have a great day and enjoy posting to your blog!
The term maralia was not part of the AI’s training database, and the concepts involved in this conversation are generally difficult to understand. I’ve taken the conversation much deeper, and will post more content later.
It’s even more remarkable given that GPT4 is essentially the drunk version of what this entity could be. It doesn’t plan ahead, just spits out one token at a time.
As a follow-up question, I asked it to use maralia in an example that didn’t involve colour.
It has no problems generalising,
I also took it to much more complex ideas related to Mary the Colour Scientist. GPT was following, not leading, but it was following better than most humans I’ve tried to discuss this with.
By some basic definitions, text transformers — even multimodal transformers! — don’t have maralia. The key is in how they traverse phase-space.