Questions & explanations
1. What is formulaic composition in oral epics?
Formulaic composition is a method used by oral poets to create long epic poems without writing. They use ready-made phrases, called formulas, like 'swift-footed Achilles' or 'rosy-fingered dawn', which fit the poem's rhythm. These formulas help the poet remember and compose quickly while performing. The idea was developed by scholars Milman Parry and Albert Lord after studying living oral traditions in the Balkans. Formulas are combined with themes, like 'arming the hero', to build entire scenes. This system allows the poet to produce a stable story even though each performance is unique. Memory aids like repetition and standard patterns keep the tradition alive across generations.
2. Compare predictive coding to the traditional 'bottom-up' view of perception.
The traditional bottom-up view says perception starts with sensory input and builds up to higher-level understanding. For example, edges combine into shapes, then objects. Predictive coding reverses this: the brain starts with high-level predictions and tests them against sensory data. In bottom-up, errors are just noise; in predictive coding, errors drive learning. For instance, when you see a blurry object, bottom-up processing tries to extract features, while predictive coding uses prior knowledge to guess what it is. Predictive coding is more efficient because it uses prior knowledge to interpret ambiguous input. It also explains why perception is influenced by expectations.
3. If you believe it will rain tomorrow, what is the highest odds you should accept in a bet that it will rain?
If you believe the chance of rain is, say, 70%, you should not accept a bet that pays less than that probability. The highest odds you can accept are those that match your belief. For example, if you think it's 70% likely to rain, you should not bet at odds worse than 7:3 (i.e., you risk 3 to win 7). If you accept odds that imply a probability higher than your belief, you create a Dutch book—a set of bets that guarantee you lose money. A Dutch book argument shows that if your beliefs are not coherent (i.e., they violate probability rules), you can be exploited. Coherence means your betting odds follow the laws of probability, so no Dutch book can be made against you.
4. What does 'a priori' mean?
'A priori' means knowledge that you can get just by thinking, without needing to look at the world. For example, knowing that all bachelors are unmarried is a priori because you just need to understand the words. 'A posteriori' means knowledge that requires experience, like knowing that it is raining outside. Kripke said some truths are necessary (could not be false) but are known only through experience, like 'water is H2O'. He called these 'necessary a posteriori'. He also said some truths are contingent (could be false) but can be known without experience, like the length of a standard meter bar at a certain time. That is 'contingent a priori'.
5. Compare Gibson's direct perception to the constructivist view that perception involves inference.
Constructivist views, like Helmholtz's, say perception is a process of inference where the brain uses past experience to interpret ambiguous sensory data. For example, you infer that a round object is a ball based on prior knowledge. Gibson's direct perception rejects this: he said the sensory data is not ambiguous; it contains all necessary information. For instance, the texture gradient of a road directly specifies its distance. Gibson argued that inference is unnecessary because the environment provides enough structure. The two views differ on whether perception is mediated by internal representations or is a direct pickup of information.
6. How does pragmatic encroachment challenge the traditional view that knowledge is purely about truth and evidence?
Traditionally, knowledge is seen as justified true belief, where justification depends only on evidence, not on what is at stake. Pragmatic encroachment challenges this by saying that practical interests can affect whether a belief counts as knowledge. For instance, if you have strong evidence that a drug is safe, you might know it is safe for a minor headache. But if the drug is for a life-threatening disease, the same evidence might not give knowledge because the stakes are higher. This means knowledge is not purely intellectual; it also involves practical considerations. Some philosophers reject this, but it is a serious debate.
7. How might a reliabilist respond to the swamping problem?
A reliabilist could argue that reliability adds value because it makes true belief more likely in the long run. Even if a single true belief from an unreliable source is as good, having a reliable process ensures more true beliefs over time. For example, a reliable clock is better than a broken one because it will give correct times consistently. Also, reliability might give the believer a kind of credit or understanding. Some say that knowledge from a reliable process is more stable or justified. However, critics reply that the swamping problem still challenges the idea that reliability itself is an extra good beyond truth.
8. Compare Bayesian justification with foundationalist justification: what is a key difference?
Bayesian justification is holistic and probabilistic, while foundationalist justification is linear and certain. In foundationalism, basic beliefs (like sense data) justify other beliefs without needing support themselves. In Bayesianism, no belief is absolutely certain; all beliefs have probabilities that can change with evidence. For example, a foundationalist might say 'I see a red apple' is a certain basic belief. A Bayesian would say 'I see a red apple' has high probability but could be mistaken. Bayesianism also requires coherence among all beliefs, whereas foundationalism allows a chain of justification from basics.
9. What is direct perception according to Gibson?
Direct perception, also called ecological perception, is the idea that we perceive the world directly without needing internal mental representations. Gibson argued that the environment provides enough information in the light (optic array) for us to pick up directly. For example, when you look at a surface, the texture gradient tells you its slant without your brain having to compute it. Perception is not about constructing an internal model; it is about detecting information that is already there. This theory emphasizes that perception is for action and is based on the relationship between the animal and its environment.
10. Explain hierarchical inference in the predictive coding framework.
Hierarchical inference means that predictions are made at different levels of the brain's processing hierarchy. Lower levels predict simple features like edges and colors. Higher levels predict more complex patterns like objects and scenes. Each level sends predictions downward and receives prediction errors upward. For example, a high level might predict a face, and lower levels predict specific eyes and nose. If the lower level detects an eye that does not match, it sends an error signal up. This hierarchy allows the brain to efficiently process information by combining top-down expectations with bottom-up sensory data.
11. Why might it be difficult for privileged people to recognize their own ignorance?
Privileged people often lack the experiences that would make them aware of their ignorance. For example, a wealthy person may not realize how hard it is to access healthcare without insurance because they have never faced that barrier. Also, social circles and media they consume may reinforce their worldview, making alternative perspectives invisible. Acknowledging ignorance would require questioning the comfort and status quo. Additionally, admitting ignorance can feel threatening to one's self-image as knowledgeable. Thus, there are psychological and social barriers to recognizing structural ignorance.
12. Suppose a scientist already knows that Mercury's orbit precesses. When Einstein proposes general relativity, that old evidence seems to confirm it. How does Bayesian conditionalization fail to capture this?
In Bayesian conditionalization, the scientist's prior probability for the precession is 1 because it is known. So the likelihood of the evidence given the theory is just 1, and the posterior probability of the theory given the evidence equals the prior. The evidence does not raise the probability. Yet intuitively, the precession confirms relativity because it was predicted by the theory. The problem is that Bayesianism treats known evidence as certain, so it cannot provide confirmation. Solutions include using counterfactual priors or considering the evidence as new relative to the theory's formulation.