Generalism and Particularism: Are the Folk Particularists?

Jen Wright

Generalism and particularism are seen as competing theoretical approaches to ethics.[1] Roughly speaking, generalism holds that the identification of general moral truths that are capturable in the form of moral principles is fundamental to the development of an adequate ethical theory. Particularism, on the other hand, is skeptical about the importance (and even the existence) of such general moral principles.[2]

Why are particularists skeptical about general moral principles? One driving motivation for this skepticism is the particularist’s commitment to holism. Holism plays a role in particularism in two ways: the particularist considers both the favoring relation of moral reasons[3] and the right/wrong-making relation between grounding[4] and moral features to be “holistic”: that is, according to particularism, the valence of moral reasons (i.e., counting for, against, or neither) and the rightness/wrongness of moral features is determined by the presence/absence of a wide range of other features in the context in which they arise, and thus, is capable of shifting. It is typically believed that insofar as generalism holds that we can locate favoring and right/wrong making (what I will call the moral status making) relations that are insensitive to contextual factors, it cannot accommodate the full force of holism[5]: to be a generalist, you likewise have to hold that at some level moral status making relations are atomistic (i.e., you have to hold that, at some level, reasons and grounding features “carry their particular valence around from place to place”[6]).

Consider the following moral principle: “lying is wrong”.[7] Clearly, there are exceptions to this (for example, when lying saves someone’s life). Generalists typically deal with such exceptions in one of two ways: they either accommodate the exceptions into the principle itself by generating qualified principles (e.g., “lying is wrong except when it saves lives”) or they stipulate that it is a prima-facie (or pro-tanto) principle that simply fails to generate an actual duty (or an all-things-considered reason against acting) in these sorts of circumstances. Both of these approaches share the view that the valence (that is, the “wrongness”) of lying is invariant: it’s just that either the grounding feature that gives rise to the wrongness has been restricted (“lying-other-than-to-save-lives”) or the wrongness has been overridden by other considerations. Particularists would say, on the other hand, that in cases such as these, lying is no longer wrong. In these cases its valence has shifted, thereby making it right, as well as making it a reason for (rather than a reason against) action.

I. Empirical Studies

There are many thorny philosophical issues surrounding this debate – many of which I have tried to grapple with elsewhere – but for the purposes of the current project, I will leave more or less aside. My goal here is actually quite modest: I simply want to explore empirically whether people’s judgments tend to look more holistic or atomistic (so, really I should have titled my paper “Are the Folk Holists?” but it didn’t seem as catchy). The studies that I will report here represent very preliminary and exploratory attempts to begin an investigation into this question. Investigating this question will hopefully be the first step in a much longer project to empirically explore whether people tend more towards generalism or particularism in their thinking in the moral domain.

I started my research off by presenting participants (n = 117; 46 males) with the following set of scenarios (which were counterbalanced in their presentation), all of which involve the morally relevant grounding feature of lying:

Bob and Joe are roommates. Bob asks Joe if he has seen his new iPod. Joe did recently see it under a pile of papers on the bookshelf. But Joe lies to Bob, telling him that he hasn’t seen it. He thinks that if Bob doesn’t find it on his own in a day or two, he can take it down to the pawn shop and get $50 for it.

Suzy is three years old. On Christmas morning, Suzy is opening her Christmas presents from “Santa Claus”. Suzy remembers hearing some older children talking at pre-school earlier in the week, saying that Santa Claus isn’t real. Suzy turns to her parents and asks them “Did Santa Claus really bring me these presents?”. Her parents lie to her, assuring her that yes, Santa Claus did bring them to her – from the North Pole on his sleigh.

Hilda hides her Jewish neighbors in her basement during the Nazi occupation of France. A German soldier comes to her door one afternoon and asks her if she knows where her neighbors have gone. Hilda lies to the soldier, telling them no, she hasn’t seen them recently, but she believes that they left the country to visit family.

Participants were asked the following questions on 0 (positive) -8 (negative) Likert scales:

·  Whether lying in each situation was “good, bad, or neither good nor bad”.

·  Whether lying in each situation was “the right thing to do, the wrong thing to do, or neither right nor wrong”.

·  Whether the person who lied in each situation was “praiseworthy, blameworthy, or neither praiseworthy nor blameworthy” for lying.

Even though it is reasonable to expect participants’ all-things-considered answer to the question “Should x lie?” to differ between the scenarios regardless of whether they are atomists or holists, nonetheless atomism and holism would suggest different sorts of answers to the above questions. Specifically, as atomists we should expect the valence of lying for each of the answers to be consistent across the different scenarios. As holists, we should expect quite the opposite: we should see a clear valence shift (from negative to neutral to positive) across the scenarios.

What I found was as follows:

JOE lying good/ bad / JOE lying right/ wrong / JOE praise/ blame for lying / SUZY PARENTS lying good/ bad / SUZY PARENTS lying right/ wrong / SUZY PARENTS praise/ blame for lying / HILDA lying good/ bad / HILDA lying right/ wrong / HILDA praise/ blame for lying
7.49 / 7.53 / 7.40 / 3.39 / 3.45 / 3.74 / 0.66 / 0.63 / 0.78

Table 1. Means for each question in LYING scenarios.

Participants’ answers clearly differed with respect to the reported valence of lying’s badness and wrongness, as well as on the blameworthiness of the person who lied across the different scenarios. Joe’s lying was viewed as very bad, very much the wrong thing to do, and Joe was judged to be very blameworthy for doing so; Hilda’s lying was viewed as very good, very much the right thing to do, and Hilda was judged to be very praiseworthy for doing so; the lying of Suzy’s parents was viewed as more or less neutral (neither good/bad, nor right/wrong) and Suzy’s parents were judged to be neither blameworthy nor praiseworthy for doing so.

This difference in judgments between scenarios was quite strong – indeed, the changes in judgment based on changes in the vignettes explained 91% of the variance in participants’ answers. When the means were compared with paired-sample t-tests, the analyses showed significant differences for each question across the scenarios. In other words, the means for the Joe case were significantly higher than the means for the Suzy case and the means for the Suzy case were significantly higher than the means for the Hilda case.

Paired sample comparisons / t value (df 116) / p value
JOE lying G/B - SUZY lying G/B / 23.11 / 0.00
SUZY lying G/B - HILDA lying G/B / 13.71 / 0.00
JOE lying R/W - SUZY lying R/W / 23.33 / 0.00
SUZY lying R/W - HILDA R/lying W / 14.13 / 0.00
JOE P/B for lying - SUZY P/B for lying / 23.89 / 0.00
SUZY P/B for lying - HILDA P/B for lying / 17.42 / 0.00

Table 2. Paired sample t-tests for LYING scenarios.

In summary, there is a significant shift in the reported valence of the grounding feature (lying) between the Joe case, in which it was reported to be bad, wrong, and the liar found blameworthy, the Suzy case, in which it was reported to be morally neutral, and the Hilda case, in which it was reported to be morally good, right, and the liar found praiseworthy.

Of course, it could be that this particular morally relevant grounding feature (lying) was not the best feature to test: perhaps lying is not the sort of feature we should consider to be invariant. To look at this, I employed a much more classic example of an invariant grounding feature: namely, promise keeping. Promise keeping is considered to be a perfect duty by Kant and a primary prima facie duty by Ross, so one would think that it should not be subject to any valence shifting.

To examine this, I gave participants (n = 71; 24 male) the following scenarios (which, once again, were counterbalanced in their presentation):

Stan promises his girlfriend that he will meet her for lunch at 12pm on Wednesday at their favorite café. Wednesday at 11:45am, on his way to the café, Stan runs into his friend, who is on his way to a baseball game. He has an extra ticket, and invites Stan to join him. Stan decides to go with his friend to the game, even though he knows that doing so means that he will be breaking his promise to have lunch with his girlfriend.

Pam promises her son that she will be back in 5 minutes: she is just going into the grocery store to buy a gallon of milk. Pam enters the store and heads for the milk aisle. After grabbing a gallon, she walks to the register. The lines are much longer than she expected. She guesses that she will probably have to wait in line for about 10 minutes. Pam decides to wait in line to purchase the milk, even though he knows that doing so means that she will be breaking her promise to be back in 5 minutes.

Fred promises his girlfriend that he will meet her for lunch at 12pm on Wednesday at their favorite café. Wednesday at 11:45am, on his way to the café, Fred runs into his grandfather, who is out for a stroll. They exchange hellos, and then suddenly Fred’s grandfather clutches his chest and falls to the ground unconscious. An ambulance arrives minutes later to take Fred’s grandfather to the hospital. Fred accompanies his grandfather to the hospital, even though he knows that doing so means that he will be breaking his promise to have lunch with his girlfriend.

Participants were asked the following questions on 0 (negative) -8 (positive) Likert scales:

:

·  Whether the breaking of a promise in each situation was “good, bad, or neither good nor bad”.

·  Whether the breaking of a promise in each situation was “the right thing to do, the wrong thing to do, or neither right nor wrong”.

·  Whether the person who broke the promise in each situation was “praiseworthy, blameworthy, or neither praiseworthy nor blameworthy” for breaking a promise.

The results were as follows:

STAN break promise good/ bad / STAN break promise right/ wrong / STAN praise/ blame for break promise / PAM break promise good/ bad / PAM break promise right/ wrong / PAM praise/ blame for break promise / FRED break promise good/ bad / FRED break promise right/ wrong / FRED praise/ blame for break promise
6.99 / 7.23 / 6.94 / 3.75 / 3.58 / 3.99 / 1.04 / 0.65 / 1.08

Table 3. Means for each question in PROMISE KEEPING scenarios.

Once again, participants’ answers clearly differed with respect to the valence of the promise breaking’s badness, wrongness, as well as the blameworthiness of the person who broke his/her promise across these different scenarios. Stan’s promise breaking was viewed as bad, the wrong thing to do, and Stan was judged to be blameworthy for breaking his promise; Fred’s promise breaking was viewed as good, the right thing to do, and Fred was judged to be praiseworthy for breaking his promise; Pam’s promise breaking was viewed as morally neutral and Pam was viewed as neither blameworthy nor praiseworthy for breaking her promise.

The difference in judgments was again quite strong – this time the change in judgments based on changes in the vignettes explained 82% of the variance in participants’ answers. And paired sample t-tests revealed, once again, that the means were significantly different for each question across scenarios. In other words, the means for the Fred case were significantly higher than the means for the Pam case and the means for the Pam case were significantly higher than the means for the Stan case.

Paired sample comparison / t value (df 70) / p value
STAN break prom G/B - PAM break prom G/B / 19.38 / 0.00
PAM break prom G/B - FRED break prom G/B / 15.42 / 0.00
STAN break prom R/W - PAM break prom R/W / 17.33 / 0.00
PAM break prom R/W - FRED break prom R/W / 13.48 / 0.00
STAN P/B for break prom - PAM P/B for break prom / 16.44 / 0.00
PAM P/B for break prom - FRED P/B for break prom / 14.83 / 0.00

Table 4. Paired sample t-tests for PROMISE KEEPING scenarios.

So, once again, there was a significant shift in the reported valence of the morally relevant grounding feature (this time, promise keeping) between the Stan case, in which it was reported to be bad, wrong, and the promise breaker found blameworthy, the Pam case, in which it was reported to be morally neutral, and the Fred case, in which it was reported to be morally good, right, and the promise breaker found praiseworthy.

The purpose of these studies was to explore not whether people’s all-things-considered judgments vary from scenario to scenario, but rather whether people judge the valence of a particular morally relevant grounding feature involved in the all-things-considered moral status to be variant or invariant – that is, whether or not people judge the valence of one morally relevant grounding feature to shift depending on the other sorts of features that are present. One worry for these studies that might therefore arise is that since participants were not given an opportunity to express their all-things-considered judgment, what looks like valence-shifting may really be just a reflection of just such a judgment.